25 files changed, 1674 insertions, 1113 deletions

diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index 69b355ae7f49..6df6d6ed74fd 100644
--- a/fs/btrfs/acl.c
+++ b/fs/btrfs/acl.c
@@ -27,7 +27,7 @@
 #include "btrfs_inode.h"
 #include "xattr.h"
 
-#ifdef CONFIG_BTRFS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 
 static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
 {
@@ -73,13 +73,13 @@ static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
         return acl;
 }
 
-static int btrfs_xattr_get_acl(struct inode *inode, int type,
-                               void *value, size_t size)
+static int btrfs_xattr_acl_get(struct dentry *dentry, const char *name,
+                void *value, size_t size, int type)
 {
         struct posix_acl *acl;
         int ret = 0;
 
-        acl = btrfs_get_acl(inode, type);
+        acl = btrfs_get_acl(dentry->d_inode, type);
 
         if (IS_ERR(acl))
                 return PTR_ERR(acl);
@@ -94,7 +94,8 @@ static int btrfs_xattr_get_acl(struct inode *inode, int type,
 /*
  * Needs to be called with fs_mutex held
  */
-static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
+static int btrfs_set_acl(struct btrfs_trans_handle *trans,
+                         struct inode *inode, struct posix_acl *acl, int type)
 {
         int ret, size = 0;
         const char *name;
@@ -111,12 +112,14 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
         switch (type) {
         case ACL_TYPE_ACCESS:
                 mode = inode->i_mode;
-                ret = posix_acl_equiv_mode(acl, &mode);
-                if (ret < 0)
-                        return ret;
-                ret = 0;
-                inode->i_mode = mode;
                 name = POSIX_ACL_XATTR_ACCESS;
+                if (acl) {
+                        ret = posix_acl_equiv_mode(acl, &mode);
+                        if (ret < 0)
+                                return ret;
+                        inode->i_mode = mode;
+                }
+                ret = 0;
                 break;
         case ACL_TYPE_DEFAULT:
                 if (!S_ISDIR(inode->i_mode))
@@ -140,8 +143,7 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
                         goto out;
         }
 
-        ret = __btrfs_setxattr(inode, name, value, size, 0);
-
+        ret = __btrfs_setxattr(trans, inode, name, value, size, 0);
 out:
         kfree(value);
 
@@ -151,10 +153,10 @@ out:
         return ret;
 }
 
-static int btrfs_xattr_set_acl(struct inode *inode, int type,
-                               const void *value, size_t size)
+static int btrfs_xattr_acl_set(struct dentry *dentry, const char *name,
+                const void *value, size_t size, int flags, int type)
 {
-        int ret = 0;
+        int ret;
         struct posix_acl *acl = NULL;
 
         if (value) {
@@ -167,38 +169,13 @@ static int btrfs_xattr_set_acl(struct inode *inode, int type,
                 }
         }
 
-        ret = btrfs_set_acl(inode, acl, type);
+        ret = btrfs_set_acl(NULL, dentry->d_inode, acl, type);
 
         posix_acl_release(acl);
 
         return ret;
 }
 
-
-static int btrfs_xattr_acl_access_get(struct inode *inode, const char *name,
-                                      void *value, size_t size)
-{
-        return btrfs_xattr_get_acl(inode, ACL_TYPE_ACCESS, value, size);
-}
-
-static int btrfs_xattr_acl_access_set(struct inode *inode, const char *name,
-                                      const void *value, size_t size, int flags)
-{
-        return btrfs_xattr_set_acl(inode, ACL_TYPE_ACCESS, value, size);
-}
-
-static int btrfs_xattr_acl_default_get(struct inode *inode, const char *name,
-                                       void *value, size_t size)
-{
-        return btrfs_xattr_get_acl(inode, ACL_TYPE_DEFAULT, value, size);
-}
-
-static int btrfs_xattr_acl_default_set(struct inode *inode, const char *name,
-                               const void *value, size_t size, int flags)
-{
-        return btrfs_xattr_set_acl(inode, ACL_TYPE_DEFAULT, value, size);
-}
-
 int btrfs_check_acl(struct inode *inode, int mask)
 {
         struct posix_acl *acl;
@@ -221,7 +198,8 @@ int btrfs_check_acl(struct inode *inode, int mask)
  * stuff has been fixed to work with that.  If the locking stuff changes, we
  * need to re-evaluate the acl locking stuff.
  */
-int btrfs_init_acl(struct inode *inode, struct inode *dir)
+int btrfs_init_acl(struct btrfs_trans_handle *trans,
+                   struct inode *inode, struct inode *dir)
 {
         struct posix_acl *acl = NULL;
         int ret = 0;
@@ -246,7 +224,8 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir)
                 mode_t mode;
 
                 if (S_ISDIR(inode->i_mode)) {
-                        ret = btrfs_set_acl(inode, acl, ACL_TYPE_DEFAULT);
+                        ret = btrfs_set_acl(trans, inode, acl,
+                                            ACL_TYPE_DEFAULT);
                         if (ret)
                                 goto failed;
                 }
@@ -261,10 +240,11 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir)
                         inode->i_mode = mode;
                         if (ret > 0) {
                                 /* we need an acl */
-                                ret = btrfs_set_acl(inode, clone,
+                                ret = btrfs_set_acl(trans, inode, clone,
                                                     ACL_TYPE_ACCESS);
                         }
                 }
+                posix_acl_release(clone);
         }
 failed:
         posix_acl_release(acl);
@@ -294,7 +274,7 @@ int btrfs_acl_chmod(struct inode *inode)
 
         ret = posix_acl_chmod_masq(clone, inode->i_mode);
         if (!ret)
-                ret = btrfs_set_acl(inode, clone, ACL_TYPE_ACCESS);
+                ret = btrfs_set_acl(NULL, inode, clone, ACL_TYPE_ACCESS);
 
         posix_acl_release(clone);
 
@@ -303,26 +283,29 @@ int btrfs_acl_chmod(struct inode *inode)
 
 struct xattr_handler btrfs_xattr_acl_default_handler = {
         .prefix = POSIX_ACL_XATTR_DEFAULT,
-        .get    = btrfs_xattr_acl_default_get,
-        .set    = btrfs_xattr_acl_default_set,
+        .flags  = ACL_TYPE_DEFAULT,
+        .get    = btrfs_xattr_acl_get,
+        .set    = btrfs_xattr_acl_set,
 };
 
 struct xattr_handler btrfs_xattr_acl_access_handler = {
         .prefix = POSIX_ACL_XATTR_ACCESS,
-        .get    = btrfs_xattr_acl_access_get,
-        .set    = btrfs_xattr_acl_access_set,
+        .flags  = ACL_TYPE_ACCESS,
+        .get    = btrfs_xattr_acl_get,
+        .set    = btrfs_xattr_acl_set,
 };
 
-#else /* CONFIG_BTRFS_POSIX_ACL */
+#else /* CONFIG_BTRFS_FS_POSIX_ACL */
 
 int btrfs_acl_chmod(struct inode *inode)
 {
         return 0;
 }
 
-int btrfs_init_acl(struct inode *inode, struct inode *dir)
+int btrfs_init_acl(struct btrfs_trans_handle *trans,
+                   struct inode *inode, struct inode *dir)
 {
         return 0;
 }
 
-#endif /* CONFIG_BTRFS_POSIX_ACL */
+#endif /* CONFIG_BTRFS_FS_POSIX_ACL */
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index c71abec0ab90..3f1f50d9d916 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -44,9 +44,6 @@ struct btrfs_inode {
          */
         struct extent_io_tree io_failure_tree;
 
-        /* held while inesrting or deleting extents from files */
-        struct mutex extent_mutex;
-
         /* held while logging the inode in tree-log.c */
         struct mutex log_mutex;
 
@@ -86,6 +83,12 @@ struct btrfs_inode {
          * transid of the trans_handle that last modified this inode
          */
         u64 last_trans;
+
+        /*
+         * log transid when this inode was last modified
+         */
+        u64 last_sub_trans;
+
         /*
          * transid that last logged this inode
          */
@@ -160,7 +163,7 @@ static inline struct btrfs_inode *BTRFS_I(struct inode *inode)
 
 static inline void btrfs_i_size_write(struct inode *inode, u64 size)
 {
-        inode->i_size = size;
+        i_size_write(inode, size);
         BTRFS_I(inode)->disk_i_size = size;
 }
 
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index ec96f3a6d536..c4bc570a396e 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -37,6 +37,11 @@ static int balance_node_right(struct btrfs_trans_handle *trans,
                               struct extent_buffer *src_buf);
 static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
                    struct btrfs_path *path, int level, int slot);
+static int setup_items_for_insert(struct btrfs_trans_handle *trans,
+                        struct btrfs_root *root, struct btrfs_path *path,
+                        struct btrfs_key *cpu_key, u32 *data_size,
+                        u32 total_data, u32 total_size, int nr);
+
 
 struct btrfs_path *btrfs_alloc_path(void)
 {
@@ -451,9 +456,8 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
                 extent_buffer_get(cow);
                 spin_unlock(&root->node_lock);
 
-                btrfs_free_extent(trans, root, buf->start, buf->len,
-                                  parent_start, root->root_key.objectid,
-                                  level, 0);
+                btrfs_free_tree_block(trans, root, buf->start, buf->len,
+                                parent_start, root->root_key.objectid, level);
                 free_extent_buffer(buf);
                 add_root_to_dirty_list(root);
         } else {
@@ -468,9 +472,8 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
                 btrfs_set_node_ptr_generation(parent, parent_slot,
                                               trans->transid);
                 btrfs_mark_buffer_dirty(parent);
-                btrfs_free_extent(trans, root, buf->start, buf->len,
-                                  parent_start, root->root_key.objectid,
-                                  level, 0);
+                btrfs_free_tree_block(trans, root, buf->start, buf->len,
+                                parent_start, root->root_key.objectid, level);
         }
         if (unlock_orig)
                 btrfs_tree_unlock(buf);
@@ -1030,8 +1033,8 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
                 btrfs_tree_unlock(mid);
                 /* once for the path */
                 free_extent_buffer(mid);
-                ret = btrfs_free_extent(trans, root, mid->start, mid->len,
-                                        0, root->root_key.objectid, level, 1);
+                ret = btrfs_free_tree_block(trans, root, mid->start, mid->len,
+                                            0, root->root_key.objectid, level);
                 /* once for the root ptr */
                 free_extent_buffer(mid);
                 return ret;
@@ -1095,10 +1098,10 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
                                        1);
                         if (wret)
                                 ret = wret;
-                        wret = btrfs_free_extent(trans, root, bytenr,
-                                                 blocksize, 0,
-                                                 root->root_key.objectid,
-                                                 level, 0);
+                        wret = btrfs_free_tree_block(trans, root,
+                                                     bytenr, blocksize, 0,
+                                                     root->root_key.objectid,
+                                                     level);
                         if (wret)
                                 ret = wret;
                 } else {
@@ -1143,9 +1146,8 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
                 wret = del_ptr(trans, root, path, level + 1, pslot);
                 if (wret)
                         ret = wret;
-                wret = btrfs_free_extent(trans, root, bytenr, blocksize,
-                                         0, root->root_key.objectid,
-                                         level, 0);
+                wret = btrfs_free_tree_block(trans, root, bytenr, blocksize,
+                                         0, root->root_key.objectid, level);
                 if (wret)
                         ret = wret;
         } else {
@@ -2997,75 +2999,85 @@ again:
         return ret;
 }
 
-/*
- * This function splits a single item into two items,
- * giving 'new_key' to the new item and splitting the
- * old one at split_offset (from the start of the item).
- *
- * The path may be released by this operation.  After
- * the split, the path is pointing to the old item.  The
- * new item is going to be in the same node as the old one.
- *
- * Note, the item being split must be smaller enough to live alone on
- * a tree block with room for one extra struct btrfs_item
- *
- * This allows us to split the item in place, keeping a lock on the
- * leaf the entire time.
- */
-int btrfs_split_item(struct btrfs_trans_handle *trans,
-                     struct btrfs_root *root,
-                     struct btrfs_path *path,
-                     struct btrfs_key *new_key,
-                     unsigned long split_offset)
+static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
+                                         struct btrfs_root *root,
+                                         struct btrfs_path *path, int ins_len)
 {
-        u32 item_size;
+        struct btrfs_key key;
         struct extent_buffer *leaf;
-        struct btrfs_key orig_key;
-        struct btrfs_item *item;
-        struct btrfs_item *new_item;
-        int ret = 0;
-        int slot;
-        u32 nritems;
-        u32 orig_offset;
-        struct btrfs_disk_key disk_key;
-        char *buf;
+        struct btrfs_file_extent_item *fi;
+        u64 extent_len = 0;
+        u32 item_size;
+        int ret;
 
         leaf = path->nodes[0];
-        btrfs_item_key_to_cpu(leaf, &orig_key, path->slots[0]);
-        if (btrfs_leaf_free_space(root, leaf) >= sizeof(struct btrfs_item))
-                goto split;
+        btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+
+        BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY &&
+               key.type != BTRFS_EXTENT_CSUM_KEY);
+
+        if (btrfs_leaf_free_space(root, leaf) >= ins_len)
+                return 0;
 
         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+        if (key.type == BTRFS_EXTENT_DATA_KEY) {
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                                    struct btrfs_file_extent_item);
+                extent_len = btrfs_file_extent_num_bytes(leaf, fi);
+        }
         btrfs_release_path(root, path);
 
-        path->search_for_split = 1;
         path->keep_locks = 1;
-
-        ret = btrfs_search_slot(trans, root, &orig_key, path, 0, 1);
+        path->search_for_split = 1;
+        ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
         path->search_for_split = 0;
+        if (ret < 0)
+                goto err;
 
+        ret = -EAGAIN;
+        leaf = path->nodes[0];
         /* if our item isn't there or got smaller, return now */
-        if (ret != 0 || item_size != btrfs_item_size_nr(path->nodes[0],
-                                                        path->slots[0])) {
-                path->keep_locks = 0;
-                return -EAGAIN;
+        if (ret > 0 || item_size != btrfs_item_size_nr(leaf, path->slots[0]))
+                goto err;
+
+        if (key.type == BTRFS_EXTENT_DATA_KEY) {
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                                    struct btrfs_file_extent_item);
+                if (extent_len != btrfs_file_extent_num_bytes(leaf, fi))
+                        goto err;
         }
 
         btrfs_set_path_blocking(path);
-        ret = split_leaf(trans, root, &orig_key, path,
-                         sizeof(struct btrfs_item), 1);
-        path->keep_locks = 0;
+        ret = split_leaf(trans, root, &key, path, ins_len, 1);
         BUG_ON(ret);
 
+        path->keep_locks = 0;
         btrfs_unlock_up_safe(path, 1);
+        return 0;
+err:
+        path->keep_locks = 0;
+        return ret;
+}
+
+static noinline int split_item(struct btrfs_trans_handle *trans,
+                               struct btrfs_root *root,
+                               struct btrfs_path *path,
+                               struct btrfs_key *new_key,
+                               unsigned long split_offset)
+{
+        struct extent_buffer *leaf;
+        struct btrfs_item *item;
+        struct btrfs_item *new_item;
+        int slot;
+        char *buf;
+        u32 nritems;
+        u32 item_size;
+        u32 orig_offset;
+        struct btrfs_disk_key disk_key;
+
         leaf = path->nodes[0];
         BUG_ON(btrfs_leaf_free_space(root, leaf) < sizeof(struct btrfs_item));
 
-split:
-        /*
-         * make sure any changes to the path from split_leaf leave it
-         * in a blocking state
-         */
         btrfs_set_path_blocking(path);
 
         item = btrfs_item_nr(leaf, path->slots[0]);
@@ -3073,19 +3085,19 @@ split:
         item_size = btrfs_item_size(leaf, item);
 
         buf = kmalloc(item_size, GFP_NOFS);
+        if (!buf)
+                return -ENOMEM;
+
         read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf,
                             path->slots[0]), item_size);
-        slot = path->slots[0] + 1;
-        leaf = path->nodes[0];
 
+        slot = path->slots[0] + 1;
         nritems = btrfs_header_nritems(leaf);
-
         if (slot != nritems) {
                 /* shift the items */
                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1),
-                              btrfs_item_nr_offset(slot),
-                              (nritems - slot) * sizeof(struct btrfs_item));
-
+                                btrfs_item_nr_offset(slot),
+                                (nritems - slot) * sizeof(struct btrfs_item));
         }
 
         btrfs_cpu_key_to_disk(&disk_key, new_key);
@@ -3113,16 +3125,81 @@ split:
                             item_size - split_offset);
         btrfs_mark_buffer_dirty(leaf);
 
-        ret = 0;
-        if (btrfs_leaf_free_space(root, leaf) < 0) {
-                btrfs_print_leaf(root, leaf);
-                BUG();
-        }
+        BUG_ON(btrfs_leaf_free_space(root, leaf) < 0);
         kfree(buf);
+        return 0;
+}
+
+/*
+ * This function splits a single item into two items,
+ * giving 'new_key' to the new item and splitting the
+ * old one at split_offset (from the start of the item).
+ *
+ * The path may be released by this operation.  After
+ * the split, the path is pointing to the old item.  The
+ * new item is going to be in the same node as the old one.
+ *
+ * Note, the item being split must be smaller enough to live alone on
+ * a tree block with room for one extra struct btrfs_item
+ *
+ * This allows us to split the item in place, keeping a lock on the
+ * leaf the entire time.
+ */
+int btrfs_split_item(struct btrfs_trans_handle *trans,
+                     struct btrfs_root *root,
+                     struct btrfs_path *path,
+                     struct btrfs_key *new_key,
+                     unsigned long split_offset)
+{
+        int ret;
+        ret = setup_leaf_for_split(trans, root, path,
+                                   sizeof(struct btrfs_item));
+        if (ret)
+                return ret;
+
+        ret = split_item(trans, root, path, new_key, split_offset);
         return ret;
 }
 
 /*
+ * This function duplicate a item, giving 'new_key' to the new item.
+ * It guarantees both items live in the same tree leaf and the new item
+ * is contiguous with the original item.
+ *
+ * This allows us to split file extent in place, keeping a lock on the
+ * leaf the entire time.
+ */
+int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
+                         struct btrfs_root *root,
+                         struct btrfs_path *path,
+                         struct btrfs_key *new_key)
+{
+        struct extent_buffer *leaf;
+        int ret;
+        u32 item_size;
+
+        leaf = path->nodes[0];
+        item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+        ret = setup_leaf_for_split(trans, root, path,
+                                   item_size + sizeof(struct btrfs_item));
+        if (ret)
+                return ret;
+
+        path->slots[0]++;
+        ret = setup_items_for_insert(trans, root, path, new_key, &item_size,
+                                     item_size, item_size +
+                                     sizeof(struct btrfs_item), 1);
+        BUG_ON(ret);
+
+        leaf = path->nodes[0];
+        memcpy_extent_buffer(leaf,
+                             btrfs_item_ptr_offset(leaf, path->slots[0]),
+                             btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
+                             item_size);
+        return 0;
+}
+
+/*
  * make the item pointed to by the path smaller.  new_size indicates
  * how small to make it, and from_end tells us if we just chop bytes
  * off the end of the item or if we shift the item to chop bytes off
@@ -3714,8 +3791,8 @@ static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans,
          */
         btrfs_unlock_up_safe(path, 0);
 
-        ret = btrfs_free_extent(trans, root, leaf->start, leaf->len,
-                                0, root->root_key.objectid, 0, 0);
+        ret = btrfs_free_tree_block(trans, root, leaf->start, leaf->len,
+                                    0, root->root_key.objectid, 0);
         return ret;
 }
 /*
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 1bb897ecdeeb..2aa8ec6a0981 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -310,6 +310,9 @@ struct btrfs_header {
 #define BTRFS_MAX_INLINE_DATA_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \
                                         sizeof(struct btrfs_item) - \
                                         sizeof(struct btrfs_file_extent_item))
+#define BTRFS_MAX_XATTR_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \
+                                 sizeof(struct btrfs_item) -\
+                                 sizeof(struct btrfs_dir_item))
 
 
 /*
@@ -859,8 +862,9 @@ struct btrfs_fs_info {
         struct mutex ordered_operations_mutex;
         struct rw_semaphore extent_commit_sem;
 
-        struct rw_semaphore subvol_sem;
+        struct rw_semaphore cleanup_work_sem;
 
+        struct rw_semaphore subvol_sem;
         struct srcu_struct subvol_srcu;
 
         struct list_head trans_list;
@@ -868,6 +872,9 @@ struct btrfs_fs_info {
         struct list_head dead_roots;
         struct list_head caching_block_groups;
 
+        spinlock_t delayed_iput_lock;
+        struct list_head delayed_iputs;
+
         atomic_t nr_async_submits;
         atomic_t async_submit_draining;
         atomic_t nr_async_bios;
@@ -1009,6 +1016,7 @@ struct btrfs_root {
         atomic_t log_writers;
         atomic_t log_commit[2];
         unsigned long log_transid;
+        unsigned long last_log_commit;
         unsigned long log_batch;
         pid_t log_start_pid;
         bool log_multiple_pids;
@@ -1033,12 +1041,12 @@ struct btrfs_root {
         int ref_cows;
         int track_dirty;
         int in_radix;
+        int clean_orphans;
 
         u64 defrag_trans_start;
         struct btrfs_key defrag_progress;
         struct btrfs_key defrag_max;
         int defrag_running;
-        int defrag_level;
         char *name;
         int in_sysfs;
 
@@ -1152,6 +1160,8 @@ struct btrfs_root {
 #define BTRFS_MOUNT_FLUSHONCOMMIT       (1 << 7)
 #define BTRFS_MOUNT_SSD_SPREAD          (1 << 8)
 #define BTRFS_MOUNT_NOSSD               (1 << 9)
+#define BTRFS_MOUNT_DISCARD             (1 << 10)
+#define BTRFS_MOUNT_FORCE_COMPRESS      (1 << 11)
 
 #define btrfs_clear_opt(o, opt)         ((o) &= ~BTRFS_MOUNT_##opt)
 #define btrfs_set_opt(o, opt)           ((o) |= BTRFS_MOUNT_##opt)
@@ -1973,6 +1983,10 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
                                         u64 parent, u64 root_objectid,
                                         struct btrfs_disk_key *key, int level,
                                         u64 hint, u64 empty_size);
+int btrfs_free_tree_block(struct btrfs_trans_handle *trans,
+                          struct btrfs_root *root,
+                          u64 bytenr, u32 blocksize,
+                          u64 parent, u64 root_objectid, int level);
 struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
                                             struct btrfs_root *root,
                                             u64 bytenr, u32 blocksize,
@@ -2087,6 +2101,10 @@ int btrfs_split_item(struct btrfs_trans_handle *trans,
                      struct btrfs_path *path,
                      struct btrfs_key *new_key,
                      unsigned long split_offset);
+int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
+                         struct btrfs_root *root,
+                         struct btrfs_path *path,
+                         struct btrfs_key *new_key);
 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
                       *root, struct btrfs_key *key, struct btrfs_path *p, int
                       ins_len, int cow);
@@ -2194,9 +2212,10 @@ int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans,
                               struct btrfs_path *path,
                               struct btrfs_dir_item *di);
 int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
-                            struct btrfs_root *root, const char *name,
-                            u16 name_len, const void *data, u16 data_len,
-                            u64 dir);
+                            struct btrfs_root *root,
+                            struct btrfs_path *path, u64 objectid,
+                            const char *name, u16 name_len,
+                            const void *data, u16 data_len);
 struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans,
                                           struct btrfs_root *root,
                                           struct btrfs_path *path, u64 dir,
@@ -2290,7 +2309,7 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
                                struct inode *inode, u64 new_size,
                                u32 min_type);
 
-int btrfs_start_delalloc_inodes(struct btrfs_root *root);
+int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput);
 int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end);
 int btrfs_writepages(struct address_space *mapping,
                      struct writeback_control *wbc);
@@ -2330,6 +2349,8 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode);
 void btrfs_orphan_cleanup(struct btrfs_root *root);
 int btrfs_cont_expand(struct inode *inode, loff_t size);
 int btrfs_invalidate_inodes(struct btrfs_root *root);
+void btrfs_add_delayed_iput(struct inode *inode);
+void btrfs_run_delayed_iputs(struct btrfs_root *root);
 extern const struct dentry_operations btrfs_dentry_operations;
 
 /* ioctl.c */
@@ -2343,12 +2364,9 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
                             int skip_pinned);
 int btrfs_check_file(struct btrfs_root *root, struct inode *inode);
 extern const struct file_operations btrfs_file_operations;
-int btrfs_drop_extents(struct btrfs_trans_handle *trans,
-                       struct btrfs_root *root, struct inode *inode,
-                       u64 start, u64 end, u64 locked_end,
-                       u64 inline_limit, u64 *hint_block, int drop_cache);
+int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
+                       u64 start, u64 end, u64 *hint_byte, int drop_cache);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
-                              struct btrfs_root *root,
                               struct inode *inode, u64 start, u64 end);
 int btrfs_release_file(struct inode *inode, struct file *file);
 
@@ -2373,12 +2391,13 @@ int btrfs_parse_options(struct btrfs_root *root, char *options);
 int btrfs_sync_fs(struct super_block *sb, int wait);
 
 /* acl.c */
-#ifdef CONFIG_BTRFS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
 int btrfs_check_acl(struct inode *inode, int mask);
 #else
 #define btrfs_check_acl NULL
 #endif
-int btrfs_init_acl(struct inode *inode, struct inode *dir);
+int btrfs_init_acl(struct btrfs_trans_handle *trans,
+                   struct inode *inode, struct inode *dir);
 int btrfs_acl_chmod(struct inode *inode);
 
 /* relocation.c */
diff --git a/fs/btrfs/dir-item.c b/fs/btrfs/dir-item.c
index f3a6075519cc..e9103b3baa49 100644
--- a/fs/btrfs/dir-item.c
+++ b/fs/btrfs/dir-item.c
@@ -68,12 +68,12 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
  * into the tree
  */
 int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
-                            struct btrfs_root *root, const char *name,
-                            u16 name_len, const void *data, u16 data_len,
-                            u64 dir)
+                            struct btrfs_root *root,
+                            struct btrfs_path *path, u64 objectid,
+                            const char *name, u16 name_len,
+                            const void *data, u16 data_len)
 {
         int ret = 0;
-        struct btrfs_path *path;
         struct btrfs_dir_item *dir_item;
         unsigned long name_ptr, data_ptr;
         struct btrfs_key key, location;
@@ -81,15 +81,11 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
         struct extent_buffer *leaf;
         u32 data_size;
 
-        key.objectid = dir;
+        BUG_ON(name_len + data_len > BTRFS_MAX_XATTR_SIZE(root));
+
+        key.objectid = objectid;
         btrfs_set_key_type(&key, BTRFS_XATTR_ITEM_KEY);
         key.offset = btrfs_name_hash(name, name_len);
-        path = btrfs_alloc_path();
-        if (!path)
-                return -ENOMEM;
-        if (name_len + data_len + sizeof(struct btrfs_dir_item) >
-            BTRFS_LEAF_DATA_SIZE(root) - sizeof(struct btrfs_item))
-                return -ENOSPC;
 
         data_size = sizeof(*dir_item) + name_len + data_len;
         dir_item = insert_with_overflow(trans, root, path, &key, data_size,
@@ -117,7 +113,6 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
         write_extent_buffer(leaf, data, data_ptr, data_len);
         btrfs_mark_buffer_dirty(path->nodes[0]);
 
-        btrfs_free_path(path);
         return ret;
 }
 
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 100551a66c46..2b59201b955c 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -892,6 +892,8 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
         root->stripesize = stripesize;
         root->ref_cows = 0;
         root->track_dirty = 0;
+        root->in_radix = 0;
+        root->clean_orphans = 0;
 
         root->fs_info = fs_info;
         root->objectid = objectid;
@@ -917,6 +919,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
         atomic_set(&root->log_writers, 0);
         root->log_batch = 0;
         root->log_transid = 0;
+        root->last_log_commit = 0;
         extent_io_tree_init(&root->dirty_log_pages,
                              fs_info->btree_inode->i_mapping, GFP_NOFS);
 
@@ -927,7 +930,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
         root->defrag_trans_start = fs_info->generation;
         init_completion(&root->kobj_unregister);
         root->defrag_running = 0;
-        root->defrag_level = 0;
         root->root_key.objectid = objectid;
         root->anon_super.s_root = NULL;
         root->anon_super.s_dev = 0;
@@ -979,12 +981,12 @@ int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans,
 
         while (1) {
                 ret = find_first_extent_bit(&log_root_tree->dirty_log_pages,
-                                    0, &start, &end, EXTENT_DIRTY);
+                                0, &start, &end, EXTENT_DIRTY | EXTENT_NEW);
                 if (ret)
                         break;
 
-                clear_extent_dirty(&log_root_tree->dirty_log_pages,
-                                   start, end, GFP_NOFS);
+                clear_extent_bits(&log_root_tree->dirty_log_pages, start, end,
+                                  EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS);
         }
         eb = fs_info->log_root_tree->node;
 
@@ -1087,6 +1089,7 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
         WARN_ON(root->log_root);
         root->log_root = log_root;
         root->log_transid = 0;
+        root->last_log_commit = 0;
         return 0;
 }
 
@@ -1208,8 +1211,10 @@ again:
         ret = radix_tree_insert(&fs_info->fs_roots_radix,
                                 (unsigned long)root->root_key.objectid,
                                 root);
-        if (ret == 0)
+        if (ret == 0) {
                 root->in_radix = 1;
+                root->clean_orphans = 1;
+        }
         spin_unlock(&fs_info->fs_roots_radix_lock);
         radix_tree_preload_end();
         if (ret) {
@@ -1223,10 +1228,6 @@ again:
         ret = btrfs_find_dead_roots(fs_info->tree_root,
                                     root->root_key.objectid);
         WARN_ON(ret);
-
-        if (!(fs_info->sb->s_flags & MS_RDONLY))
-                btrfs_orphan_cleanup(root);
-
         return root;
 fail:
         free_fs_root(root);
@@ -1475,6 +1476,7 @@ static int cleaner_kthread(void *arg)
 
                 if (!(root->fs_info->sb->s_flags & MS_RDONLY) &&
                     mutex_trylock(&root->fs_info->cleaner_mutex)) {
+                        btrfs_run_delayed_iputs(root);
                         btrfs_clean_old_snapshots(root);
                         mutex_unlock(&root->fs_info->cleaner_mutex);
                 }
@@ -1604,6 +1606,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
         INIT_LIST_HEAD(&fs_info->trans_list);
         INIT_LIST_HEAD(&fs_info->dead_roots);
+        INIT_LIST_HEAD(&fs_info->delayed_iputs);
         INIT_LIST_HEAD(&fs_info->hashers);
         INIT_LIST_HEAD(&fs_info->delalloc_inodes);
         INIT_LIST_HEAD(&fs_info->ordered_operations);
@@ -1612,6 +1615,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         spin_lock_init(&fs_info->new_trans_lock);
         spin_lock_init(&fs_info->ref_cache_lock);
         spin_lock_init(&fs_info->fs_roots_radix_lock);
+        spin_lock_init(&fs_info->delayed_iput_lock);
 
         init_completion(&fs_info->kobj_unregister);
         fs_info->tree_root = tree_root;
@@ -1687,6 +1691,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         mutex_init(&fs_info->cleaner_mutex);
         mutex_init(&fs_info->volume_mutex);
         init_rwsem(&fs_info->extent_commit_sem);
+        init_rwsem(&fs_info->cleanup_work_sem);
         init_rwsem(&fs_info->subvol_sem);
 
         btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
@@ -1977,7 +1982,12 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 
         if (!(sb->s_flags & MS_RDONLY)) {
                 ret = btrfs_recover_relocation(tree_root);
-                BUG_ON(ret);
+                if (ret < 0) {
+                        printk(KERN_WARNING
+                               "btrfs: failed to recover relocation\n");
+                        err = -EINVAL;
+                        goto fail_trans_kthread;
+                }
         }
 
         location.objectid = BTRFS_FS_TREE_OBJECTID;
@@ -1988,6 +1998,12 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         if (!fs_info->fs_root)
                 goto fail_trans_kthread;
 
+        if (!(sb->s_flags & MS_RDONLY)) {
+                down_read(&fs_info->cleanup_work_sem);
+                btrfs_orphan_cleanup(fs_info->fs_root);
+                up_read(&fs_info->cleanup_work_sem);
+        }
+
         return tree_root;
 
 fail_trans_kthread:
@@ -2384,8 +2400,14 @@ int btrfs_commit_super(struct btrfs_root *root)
         int ret;
 
         mutex_lock(&root->fs_info->cleaner_mutex);
+        btrfs_run_delayed_iputs(root);
         btrfs_clean_old_snapshots(root);
         mutex_unlock(&root->fs_info->cleaner_mutex);
+
+        /* wait until ongoing cleanup work done */
+        down_write(&root->fs_info->cleanup_work_sem);
+        up_write(&root->fs_info->cleanup_work_sem);
+
         trans = btrfs_start_transaction(root, 1);
         ret = btrfs_commit_transaction(trans, root);
         BUG_ON(ret);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index d0c4d584efad..559f72489b3b 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -83,6 +83,17 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
         return (cache->flags & bits) == bits;
 }
 
+void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
+{
+        atomic_inc(&cache->count);
+}
+
+void btrfs_put_block_group(struct btrfs_block_group_cache *cache)
+{
+        if (atomic_dec_and_test(&cache->count))
+                kfree(cache);
+}
+
 /*
  * this adds the block group to the fs_info rb tree for the block group
  * cache
@@ -156,7 +167,7 @@ block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr,
                 }
         }
         if (ret)
-                atomic_inc(&ret->count);
+                btrfs_get_block_group(ret);
         spin_unlock(&info->block_group_cache_lock);
 
         return ret;
@@ -195,6 +206,14 @@ static int exclude_super_stripes(struct btrfs_root *root,
         int stripe_len;
         int i, nr, ret;
 
+        if (cache->key.objectid < BTRFS_SUPER_INFO_OFFSET) {
+                stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->key.objectid;
+                cache->bytes_super += stripe_len;
+                ret = add_excluded_extent(root, cache->key.objectid,
+                                          stripe_len);
+                BUG_ON(ret);
+        }
+
         for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
                 bytenr = btrfs_sb_offset(i);
                 ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
@@ -255,7 +274,7 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
                 if (ret)
                         break;
 
-                if (extent_start == start) {
+                if (extent_start <= start) {
                         start = extent_end + 1;
                 } else if (extent_start > start && extent_start < end) {
                         size = extent_start - start;
@@ -399,6 +418,8 @@ err:
 
         put_caching_control(caching_ctl);
         atomic_dec(&block_group->space_info->caching_threads);
+        btrfs_put_block_group(block_group);
+
         return 0;
 }
 
@@ -439,6 +460,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache)
         up_write(&fs_info->extent_commit_sem);
 
         atomic_inc(&cache->space_info->caching_threads);
+        btrfs_get_block_group(cache);
 
         tsk = kthread_run(caching_kthread, cache, "btrfs-cache-%llu\n",
                           cache->key.objectid);
@@ -478,12 +500,6 @@ struct btrfs_block_group_cache *btrfs_lookup_block_group(
         return cache;
 }
 
-void btrfs_put_block_group(struct btrfs_block_group_cache *cache)
-{
-        if (atomic_dec_and_test(&cache->count))
-                kfree(cache);
-}
-
 static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
                                                   u64 flags)
 {
@@ -1568,23 +1584,23 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans,
         return ret;
 }
 
-#ifdef BIO_RW_DISCARD
 static void btrfs_issue_discard(struct block_device *bdev,
                                 u64 start, u64 len)
 {
         blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL,
                              DISCARD_FL_BARRIER);
 }
-#endif
 
 static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
                                 u64 num_bytes)
 {
-#ifdef BIO_RW_DISCARD
         int ret;
         u64 map_length = num_bytes;
         struct btrfs_multi_bio *multi = NULL;
 
+        if (!btrfs_test_opt(root, DISCARD))
+                return 0;
+
         /* Tell the block device(s) that the sectors can be discarded */
         ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
                               bytenr, &map_length, &multi, 0);
@@ -1604,9 +1620,6 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
         }
 
         return ret;
-#else
-        return 0;
-#endif
 }
 
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
@@ -2577,7 +2590,7 @@ next_block_group(struct btrfs_root *root,
         if (node) {
                 cache = rb_entry(node, struct btrfs_block_group_cache,
                                  cache_node);
-                atomic_inc(&cache->count);
+                btrfs_get_block_group(cache);
         } else
                 cache = NULL;
         spin_unlock(&root->fs_info->block_group_cache_lock);
@@ -2883,9 +2896,9 @@ static noinline void flush_delalloc_async(struct btrfs_work *work)
         root = async->root;
         info = async->info;
 
-        btrfs_start_delalloc_inodes(root);
+        btrfs_start_delalloc_inodes(root, 0);
         wake_up(&info->flush_wait);
-        btrfs_wait_ordered_extents(root, 0);
+        btrfs_wait_ordered_extents(root, 0, 0);
 
         spin_lock(&info->lock);
         info->flushing = 0;
@@ -2959,8 +2972,8 @@ static void flush_delalloc(struct btrfs_root *root,
         return;
 
 flush:
-        btrfs_start_delalloc_inodes(root);
-        btrfs_wait_ordered_extents(root, 0);
+        btrfs_start_delalloc_inodes(root, 0);
+        btrfs_wait_ordered_extents(root, 0, 0);
 
         spin_lock(&info->lock);
         info->flushing = 0;
@@ -2980,10 +2993,10 @@ static int maybe_allocate_chunk(struct btrfs_root *root,
 
         free_space = btrfs_super_total_bytes(disk_super);
         /*
-         * we allow the metadata to grow to a max of either 5gb or 5% of the
+         * we allow the metadata to grow to a max of either 10gb or 5% of the
          * space in the volume.
          */
-        min_metadata = min((u64)5 * 1024 * 1024 * 1024,
+        min_metadata = min((u64)10 * 1024 * 1024 * 1024,
                              div64_u64(free_space * 5, 100));
         if (info->total_bytes >= min_metadata) {
                 spin_unlock(&info->lock);
@@ -3457,14 +3470,6 @@ static int update_block_group(struct btrfs_trans_handle *trans,
         else
                 old_val -= num_bytes;
         btrfs_set_super_bytes_used(&info->super_copy, old_val);
-
-        /* block accounting for root item */
-        old_val = btrfs_root_used(&root->root_item);
-        if (alloc)
-                old_val += num_bytes;
-        else
-                old_val -= num_bytes;
-        btrfs_set_root_used(&root->root_item, old_val);
         spin_unlock(&info->delalloc_lock);
 
         while (total) {
@@ -3690,6 +3695,14 @@ static int pin_down_bytes(struct btrfs_trans_handle *trans,
         if (is_data)
                 goto pinit;
 
+        /*
+         * discard is sloooow, and so triggering discards on
+         * individual btree blocks isn't a good plan.  Just
+         * pin everything in discard mode.
+         */
+        if (btrfs_test_opt(root, DISCARD))
+                goto pinit;
+
         buf = btrfs_find_tree_block(root, bytenr, num_bytes);
         if (!buf)
                 goto pinit;
@@ -4044,6 +4057,21 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans,
         return ret;
 }
 
+int btrfs_free_tree_block(struct btrfs_trans_handle *trans,
+                          struct btrfs_root *root,
+                          u64 bytenr, u32 blocksize,
+                          u64 parent, u64 root_objectid, int level)
+{
+        u64 used;
+        spin_lock(&root->node_lock);
+        used = btrfs_root_used(&root->root_item) - blocksize;
+        btrfs_set_root_used(&root->root_item, used);
+        spin_unlock(&root->node_lock);
+
+        return btrfs_free_extent(trans, root, bytenr, blocksize,
+                                 parent, root_objectid, level, 0);
+}
+
 static u64 stripe_align(struct btrfs_root *root, u64 val)
 {
         u64 mask = ((u64)root->stripesize - 1);
@@ -4097,7 +4125,7 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
 }
 
 enum btrfs_loop_type {
-        LOOP_CACHED_ONLY = 0,
+        LOOP_FIND_IDEAL = 0,
         LOOP_CACHING_NOWAIT = 1,
         LOOP_CACHING_WAIT = 2,
         LOOP_ALLOC_CHUNK = 3,
@@ -4126,12 +4154,15 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
         struct btrfs_block_group_cache *block_group = NULL;
         int empty_cluster = 2 * 1024 * 1024;
         int allowed_chunk_alloc = 0;
+        int done_chunk_alloc = 0;
         struct btrfs_space_info *space_info;
         int last_ptr_loop = 0;
         int loop = 0;
         bool found_uncached_bg = false;
         bool failed_cluster_refill = false;
         bool failed_alloc = false;
+        u64 ideal_cache_percent = 0;
+        u64 ideal_cache_offset = 0;
 
         WARN_ON(num_bytes < root->sectorsize);
         btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
@@ -4167,14 +4198,19 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
                 empty_cluster = 0;
 
         if (search_start == hint_byte) {
+ideal_cache:
                 block_group = btrfs_lookup_block_group(root->fs_info,
                                                        search_start);
                 /*
                  * we don't want to use the block group if it doesn't match our
                  * allocation bits, or if its not cached.
+                 *
+                 * However if we are re-searching with an ideal block group
+                 * picked out then we don't care that the block group is cached.
                  */
                 if (block_group && block_group_bits(block_group, data) &&
-                    block_group_cache_done(block_group)) {
+                    (block_group->cached != BTRFS_CACHE_NO ||
+                     search_start == ideal_cache_offset)) {
                         down_read(&space_info->groups_sem);
                         if (list_empty(&block_group->list) ||
                             block_group->ro) {
@@ -4186,46 +4222,63 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
                                  */
                                 btrfs_put_block_group(block_group);
                                 up_read(&space_info->groups_sem);
-                        } else
+                        } else {
                                 goto have_block_group;
+                        }
                 } else if (block_group) {
                         btrfs_put_block_group(block_group);
                 }
         }
-
 search:
         down_read(&space_info->groups_sem);
         list_for_each_entry(block_group, &space_info->block_groups, list) {
                 u64 offset;
                 int cached;
 
-                atomic_inc(&block_group->count);
+                btrfs_get_block_group(block_group);
                 search_start = block_group->key.objectid;
 
 have_block_group:
                 if (unlikely(block_group->cached == BTRFS_CACHE_NO)) {
+                        u64 free_percent;
+
+                        free_percent = btrfs_block_group_used(&block_group->item);
+                        free_percent *= 100;
+                        free_percent = div64_u64(free_percent,
+                                                 block_group->key.offset);
+                        free_percent = 100 - free_percent;
+                        if (free_percent > ideal_cache_percent &&
+                            likely(!block_group->ro)) {
+                                ideal_cache_offset = block_group->key.objectid;
+                                ideal_cache_percent = free_percent;
+                        }
+
                         /*
-                         * we want to start caching kthreads, but not too many
-                         * right off the bat so we don't overwhelm the system,
-                         * so only start them if there are less than 2 and we're
-                         * in the initial allocation phase.
+                         * We only want to start kthread caching if we are at
+                         * the point where we will wait for caching to make
+                         * progress, or if our ideal search is over and we've
+                         * found somebody to start caching.
                          */
                         if (loop > LOOP_CACHING_NOWAIT ||
-                            atomic_read(&space_info->caching_threads) < 2) {
+                            (loop > LOOP_FIND_IDEAL &&
+                             atomic_read(&space_info->caching_threads) < 2)) {
                                 ret = cache_block_group(block_group);
                                 BUG_ON(ret);
                         }
-                }
-
-                cached = block_group_cache_done(block_group);
-                if (unlikely(!cached)) {
                         found_uncached_bg = true;
 
-                        /* if we only want cached bgs, loop */
-                        if (loop == LOOP_CACHED_ONLY)
+                        /*
+                         * If loop is set for cached only, try the next block
+                         * group.
+                         */
+                        if (loop == LOOP_FIND_IDEAL)
                                 goto loop;
                 }
 
+                cached = block_group_cache_done(block_group);
+                if (unlikely(!cached))
+                        found_uncached_bg = true;
+
                 if (unlikely(block_group->ro))
                         goto loop;
 
@@ -4270,7 +4323,7 @@ have_block_group:
 
                                 btrfs_put_block_group(block_group);
                                 block_group = last_ptr->block_group;
-                                atomic_inc(&block_group->count);
+                                btrfs_get_block_group(block_group);
                                 spin_unlock(&last_ptr->lock);
                                 spin_unlock(&last_ptr->refill_lock);
 
@@ -4405,9 +4458,11 @@ loop:
         }
         up_read(&space_info->groups_sem);
 
-        /* LOOP_CACHED_ONLY, only search fully cached block groups
-         * LOOP_CACHING_NOWAIT, search partially cached block groups, but
-         *                      dont wait foR them to finish caching
+        /* LOOP_FIND_IDEAL, only search caching/cached bg's, and don't wait for
+         *                      for them to make caching progress.  Also
+         *                      determine the best possible bg to cache
+         * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking
+         *                      caching kthreads as we move along
          * LOOP_CACHING_WAIT, search everything, and wait if our bg is caching
          * LOOP_ALLOC_CHUNK, force a chunk allocation and try again
          * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try
@@ -4416,12 +4471,47 @@ loop:
         if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE &&
             (found_uncached_bg || empty_size || empty_cluster ||
              allowed_chunk_alloc)) {
-                if (found_uncached_bg) {
+                if (loop == LOOP_FIND_IDEAL && found_uncached_bg) {
                         found_uncached_bg = false;
-                        if (loop < LOOP_CACHING_WAIT) {
-                                loop++;
+                        loop++;
+                        if (!ideal_cache_percent &&
+                            atomic_read(&space_info->caching_threads))
                                 goto search;
-                        }
+
+                        /*
+                         * 1 of the following 2 things have happened so far
+                         *
+                         * 1) We found an ideal block group for caching that
+                         * is mostly full and will cache quickly, so we might
+                         * as well wait for it.
+                         *
+                         * 2) We searched for cached only and we didn't find
+                         * anything, and we didn't start any caching kthreads
+                         * either, so chances are we will loop through and
+                         * start a couple caching kthreads, and then come back
+                         * around and just wait for them.  This will be slower
+                         * because we will have 2 caching kthreads reading at
+                         * the same time when we could have just started one
+                         * and waited for it to get far enough to give us an
+                         * allocation, so go ahead and go to the wait caching
+                         * loop.
+                         */
+                        loop = LOOP_CACHING_WAIT;
+                        search_start = ideal_cache_offset;
+                        ideal_cache_percent = 0;
+                        goto ideal_cache;
+                } else if (loop == LOOP_FIND_IDEAL) {
+                        /*
+                         * Didn't find a uncached bg, wait on anything we find
+                         * next.
+                         */
+                        loop = LOOP_CACHING_WAIT;
+                        goto search;
+                }
+
+                if (loop < LOOP_CACHING_WAIT) {
+                        loop++;
+                        goto search;
                 }
 
                 if (loop == LOOP_ALLOC_CHUNK) {
@@ -4433,7 +4523,8 @@ loop:
                         ret = do_chunk_alloc(trans, root, num_bytes +
                                              2 * 1024 * 1024, data, 1);
                         allowed_chunk_alloc = 0;
-                } else {
+                        done_chunk_alloc = 1;
+                } else if (!done_chunk_alloc) {
                         space_info->force_alloc = 1;
                 }
 
@@ -4510,7 +4601,6 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 {
         int ret;
         u64 search_start = 0;
-        struct btrfs_fs_info *info = root->fs_info;
 
         data = btrfs_get_alloc_profile(root, data);
 again:
@@ -4518,17 +4608,9 @@ again:
          * the only place that sets empty_size is btrfs_realloc_node, which
          * is not called recursively on allocations
          */
-        if (empty_size || root->ref_cows) {
-                if (!(data & BTRFS_BLOCK_GROUP_METADATA)) {
-                        ret = do_chunk_alloc(trans, root->fs_info->extent_root,
-                                     2 * 1024 * 1024,
-                                     BTRFS_BLOCK_GROUP_METADATA |
-                                     (info->metadata_alloc_profile &
-                                      info->avail_metadata_alloc_bits), 0);
-                }
+        if (empty_size || root->ref_cows)
                 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
                                      num_bytes + 2 * 1024 * 1024, data, 0);
-        }
 
         WARN_ON(num_bytes < root->sectorsize);
         ret = find_free_extent(trans, root, num_bytes, empty_size,
@@ -4829,6 +4911,14 @@ static int alloc_tree_block(struct btrfs_trans_handle *trans,
                                         extent_op);
                 BUG_ON(ret);
         }
+
+        if (root_objectid == root->root_key.objectid) {
+                u64 used;
+                spin_lock(&root->node_lock);
+                used = btrfs_root_used(&root->root_item) + num_bytes;
+                btrfs_set_root_used(&root->root_item, used);
+                spin_unlock(&root->node_lock);
+        }
         return ret;
 }
 
@@ -4851,8 +4941,16 @@ struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
         btrfs_set_buffer_uptodate(buf);
 
         if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
-                set_extent_dirty(&root->dirty_log_pages, buf->start,
-                         buf->start + buf->len - 1, GFP_NOFS);
+                /*
+                 * we allow two log transactions at a time, use different
+                 * EXENT bit to differentiate dirty pages.
+                 */
+                if (root->log_transid % 2 == 0)
+                        set_extent_dirty(&root->dirty_log_pages, buf->start,
+                                        buf->start + buf->len - 1, GFP_NOFS);
+                else
+                        set_extent_new(&root->dirty_log_pages, buf->start,
+                                        buf->start + buf->len - 1, GFP_NOFS);
         } else {
                 set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
                          buf->start + buf->len - 1, GFP_NOFS);
@@ -5304,10 +5402,6 @@ static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
         int ret;
 
         while (level >= 0) {
-                if (path->slots[level] >=
-                    btrfs_header_nritems(path->nodes[level]))
-                        break;
-
                 ret = walk_down_proc(trans, root, path, wc, lookup_info);
                 if (ret > 0)
                         break;
@@ -5315,6 +5409,10 @@ static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
                 if (level == 0)
                         break;
 
+                if (path->slots[level] >=
+                    btrfs_header_nritems(path->nodes[level]))
+                        break;
+
                 ret = do_walk_down(trans, root, path, wc, &lookup_info);
                 if (ret > 0) {
                         path->slots[level]++;
@@ -7305,9 +7403,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
                         wait_block_group_cache_done(block_group);
 
                 btrfs_remove_free_space_cache(block_group);
-
-                WARN_ON(atomic_read(&block_group->count) != 1);
-                kfree(block_group);
+                btrfs_put_block_group(block_group);
 
                 spin_lock(&info->block_group_cache_lock);
         }
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 96577e8bf9fd..b177ed319612 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -3165,10 +3165,9 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
                 spin_unlock(&tree->buffer_lock);
                 goto free_eb;
         }
-        spin_unlock(&tree->buffer_lock);
-
         /* add one reference for the tree */
         atomic_inc(&eb->refs);
+        spin_unlock(&tree->buffer_lock);
         return eb;
 
 free_eb:
diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c
index 2c726b7b9faa..428fcac45f90 100644
--- a/fs/btrfs/extent_map.c
+++ b/fs/btrfs/extent_map.c
@@ -155,20 +155,6 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset,
         return NULL;
 }
 
-/*
- * look for an offset in the tree, and if it can't be found, return
- * the first offset we can find smaller than 'offset'.
- */
-static inline struct rb_node *tree_search(struct rb_root *root, u64 offset)
-{
-        struct rb_node *prev;
-        struct rb_node *ret;
-        ret = __tree_search(root, offset, &prev, NULL);
-        if (!ret)
-                return prev;
-        return ret;
-}
-
 /* check to see if two extent_map structs are adjacent and safe to merge */
 static int mergable_maps(struct extent_map *prev, struct extent_map *next)
 {
@@ -208,7 +194,7 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
         write_lock(&tree->lock);
         em = lookup_extent_mapping(tree, start, len);
 
-        WARN_ON(em->start != start || !em);
+        WARN_ON(!em || em->start != start);
 
         if (!em)
                 goto out;
@@ -256,7 +242,7 @@ out:
  * Insert @em into @tree or perform a simple forward/backward merge with
  * existing mappings.  The extent_map struct passed in will be inserted
  * into the tree directly, with an additional reference taken, or a
- * reference dropped if the merge attempt was sucessfull.
+ * reference dropped if the merge attempt was successfull.
  */
 int add_extent_mapping(struct extent_map_tree *tree,
                        struct extent_map *em)
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 2d623aa0625f..6ed434ac037f 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -179,18 +179,14 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
                 }
                 flags = em->flags;
                 if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
-                        if (em->start <= start &&
-                            (!testend || em->start + em->len >= start + len)) {
+                        if (testend && em->start + em->len >= start + len) {
                                 free_extent_map(em);
                                 write_unlock(&em_tree->lock);
                                 break;
                         }
-                        if (start < em->start) {
-                                len = em->start - start;
-                        } else {
+                        start = em->start + em->len;
+                        if (testend)
                                 len = start + len - (em->start + em->len);
-                                start = em->start + em->len;
-                        }
                         free_extent_map(em);
                         write_unlock(&em_tree->lock);
                         continue;
@@ -265,324 +261,253 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
  * If an extent intersects the range but is not entirely inside the range
  * it is either truncated or split.  Anything entirely inside the range
  * is deleted from the tree.
- *
- * inline_limit is used to tell this code which offsets in the file to keep
- * if they contain inline extents.
  */
-noinline int btrfs_drop_extents(struct btrfs_trans_handle *trans,
-                       struct btrfs_root *root, struct inode *inode,
-                       u64 start, u64 end, u64 locked_end,
-                       u64 inline_limit, u64 *hint_byte, int drop_cache)
+int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
+                       u64 start, u64 end, u64 *hint_byte, int drop_cache)
 {
-        u64 extent_end = 0;
-        u64 search_start = start;
-        u64 ram_bytes = 0;
-        u64 disk_bytenr = 0;
-        u64 orig_locked_end = locked_end;
-        u8 compression;
-        u8 encryption;
-        u16 other_encoding = 0;
+        struct btrfs_root *root = BTRFS_I(inode)->root;
         struct extent_buffer *leaf;
-        struct btrfs_file_extent_item *extent;
+        struct btrfs_file_extent_item *fi;
         struct btrfs_path *path;
         struct btrfs_key key;
-        struct btrfs_file_extent_item old;
-        int keep;
-        int slot;
-        int bookend;
-        int found_type = 0;
-        int found_extent;
-        int found_inline;
+        struct btrfs_key new_key;
+        u64 search_start = start;
+        u64 disk_bytenr = 0;
+        u64 num_bytes = 0;
+        u64 extent_offset = 0;
+        u64 extent_end = 0;
+        int del_nr = 0;
+        int del_slot = 0;
+        int extent_type;
         int recow;
         int ret;
 
-        inline_limit = 0;
         if (drop_cache)
                 btrfs_drop_extent_cache(inode, start, end - 1, 0);
 
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
+
         while (1) {
                 recow = 0;
-                btrfs_release_path(root, path);
                 ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino,
                                                search_start, -1);
                 if (ret < 0)
-                        goto out;
-                if (ret > 0) {
-                        if (path->slots[0] == 0) {
-                                ret = 0;
-                                goto out;
-                        }
-                        path->slots[0]--;
+                        break;
+                if (ret > 0 && path->slots[0] > 0 && search_start == start) {
+                        leaf = path->nodes[0];
+                        btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1);
+                        if (key.objectid == inode->i_ino &&
+                            key.type == BTRFS_EXTENT_DATA_KEY)
+                                path->slots[0]--;
                 }
+                ret = 0;
 next_slot:
-                keep = 0;
-                bookend = 0;
-                found_extent = 0;
-                found_inline = 0;
-                compression = 0;
-                encryption = 0;
-                extent = NULL;
                 leaf = path->nodes[0];
-                slot = path->slots[0];
-                ret = 0;
-                btrfs_item_key_to_cpu(leaf, &key, slot);
-                if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY &&
-                    key.offset >= end) {
-                        goto out;
-                }
-                if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
-                    key.objectid != inode->i_ino) {
-                        goto out;
-                }
-                if (recow) {
-                        search_start = max(key.offset, start);
-                        continue;
-                }
-                if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
-                        extent = btrfs_item_ptr(leaf, slot,
-                                                struct btrfs_file_extent_item);
-                        found_type = btrfs_file_extent_type(leaf, extent);
-                        compression = btrfs_file_extent_compression(leaf,
-                                                                    extent);
-                        encryption = btrfs_file_extent_encryption(leaf,
-                                                                  extent);
-                        other_encoding = btrfs_file_extent_other_encoding(leaf,
-                                                                  extent);
-                        if (found_type == BTRFS_FILE_EXTENT_REG ||
-                            found_type == BTRFS_FILE_EXTENT_PREALLOC) {
-                                extent_end =
-                                     btrfs_file_extent_disk_bytenr(leaf,
-                                                                   extent);
-                                if (extent_end)
-                                        *hint_byte = extent_end;
-
-                                extent_end = key.offset +
-                                     btrfs_file_extent_num_bytes(leaf, extent);
-                                ram_bytes = btrfs_file_extent_ram_bytes(leaf,
-                                                                extent);
-                                found_extent = 1;
-                        } else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
-                                found_inline = 1;
-                                extent_end = key.offset +
-                                     btrfs_file_extent_inline_len(leaf, extent);
+                if (path->slots[0] >= btrfs_header_nritems(leaf)) {
+                        BUG_ON(del_nr > 0);
+                        ret = btrfs_next_leaf(root, path);
+                        if (ret < 0)
+                                break;
+                        if (ret > 0) {
+                                ret = 0;
+                                break;
                         }
+                        leaf = path->nodes[0];
+                        recow = 1;
+                }
+
+                btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+                if (key.objectid > inode->i_ino ||
+                    key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end)
+                        break;
+
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                                    struct btrfs_file_extent_item);
+                extent_type = btrfs_file_extent_type(leaf, fi);
+
+                if (extent_type == BTRFS_FILE_EXTENT_REG ||
+                    extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
+                        disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
+                        num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
+                        extent_offset = btrfs_file_extent_offset(leaf, fi);
+                        extent_end = key.offset +
+                                btrfs_file_extent_num_bytes(leaf, fi);
+                } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+                        extent_end = key.offset +
+                                btrfs_file_extent_inline_len(leaf, fi);
                 } else {
+                        WARN_ON(1);
                         extent_end = search_start;
                 }
 
-                /* we found nothing we can drop */
-                if ((!found_extent && !found_inline) ||
-                    search_start >= extent_end) {
-                        int nextret;
-                        u32 nritems;
-                        nritems = btrfs_header_nritems(leaf);
-                        if (slot >= nritems - 1) {
-                                nextret = btrfs_next_leaf(root, path);
-                                if (nextret)
-                                        goto out;
-                                recow = 1;
-                        } else {
-                                path->slots[0]++;
-                        }
+                if (extent_end <= search_start) {
+                        path->slots[0]++;
                         goto next_slot;
                 }
 
-                if (end <= extent_end && start >= key.offset && found_inline)
-                        *hint_byte = EXTENT_MAP_INLINE;
-
-                if (found_extent) {
-                        read_extent_buffer(leaf, &old, (unsigned long)extent,
-                                           sizeof(old));
-                }
-
-                if (end < extent_end && end >= key.offset) {
-                        bookend = 1;
-                        if (found_inline && start <= key.offset)
-                                keep = 1;
+                search_start = max(key.offset, start);
+                if (recow) {
+                        btrfs_release_path(root, path);
+                        continue;
                 }
 
-                if (bookend && found_extent) {
-                        if (locked_end < extent_end) {
-                                ret = try_lock_extent(&BTRFS_I(inode)->io_tree,
-                                                locked_end, extent_end - 1,
-                                                GFP_NOFS);
-                                if (!ret) {
-                                        btrfs_release_path(root, path);
-                                        lock_extent(&BTRFS_I(inode)->io_tree,
-                                                locked_end, extent_end - 1,
-                                                GFP_NOFS);
-                                        locked_end = extent_end;
-                                        continue;
-                                }
-                                locked_end = extent_end;
+                /*
+                 *     | - range to drop - |
+                 *  | -------- extent -------- |
+                 */
+                if (start > key.offset && end < extent_end) {
+                        BUG_ON(del_nr > 0);
+                        BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);
+
+                        memcpy(&new_key, &key, sizeof(new_key));
+                        new_key.offset = start;
+                        ret = btrfs_duplicate_item(trans, root, path,
+                                                   &new_key);
+                        if (ret == -EAGAIN) {
+                                btrfs_release_path(root, path);
+                                continue;
                         }
-                        disk_bytenr = le64_to_cpu(old.disk_bytenr);
-                        if (disk_bytenr != 0) {
+                        if (ret < 0)
+                                break;
+
+                        leaf = path->nodes[0];
+                        fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
+                                            struct btrfs_file_extent_item);
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        start - key.offset);
+
+                        fi = btrfs_item_ptr(leaf, path->slots[0],
+                                            struct btrfs_file_extent_item);
+
+                        extent_offset += start - key.offset;
+                        btrfs_set_file_extent_offset(leaf, fi, extent_offset);
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        extent_end - start);
+                        btrfs_mark_buffer_dirty(leaf);
+
+                        if (disk_bytenr > 0) {
                                 ret = btrfs_inc_extent_ref(trans, root,
-                                           disk_bytenr,
-                                           le64_to_cpu(old.disk_num_bytes), 0,
-                                           root->root_key.objectid,
-                                           key.objectid, key.offset -
-                                           le64_to_cpu(old.offset));
+                                                disk_bytenr, num_bytes, 0,
+                                                root->root_key.objectid,
+                                                new_key.objectid,
+                                                start - extent_offset);
                                 BUG_ON(ret);
+                                *hint_byte = disk_bytenr;
                         }
+                        key.offset = start;
                 }
+                /*
+                 *  | ---- range to drop ----- |
+                 *      | -------- extent -------- |
+                 */
+                if (start <= key.offset && end < extent_end) {
+                        BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);
+
+                        memcpy(&new_key, &key, sizeof(new_key));
+                        new_key.offset = end;
+                        btrfs_set_item_key_safe(trans, root, path, &new_key);
 
-                if (found_inline) {
-                        u64 mask = root->sectorsize - 1;
-                        search_start = (extent_end + mask) & ~mask;
-                } else
-                        search_start = extent_end;
-
-                /* truncate existing extent */
-                if (start > key.offset) {
-                        u64 new_num;
-                        u64 old_num;
-                        keep = 1;
-                        WARN_ON(start & (root->sectorsize - 1));
-                        if (found_extent) {
-                                new_num = start - key.offset;
-                                old_num = btrfs_file_extent_num_bytes(leaf,
-                                                                      extent);
-                                *hint_byte =
-                                        btrfs_file_extent_disk_bytenr(leaf,
-                                                                      extent);
-                                if (btrfs_file_extent_disk_bytenr(leaf,
-                                                                  extent)) {
-                                        inode_sub_bytes(inode, old_num -
-                                                        new_num);
-                                }
-                                btrfs_set_file_extent_num_bytes(leaf,
-                                                        extent, new_num);
-                                btrfs_mark_buffer_dirty(leaf);
-                        } else if (key.offset < inline_limit &&
-                                   (end > extent_end) &&
-                                   (inline_limit < extent_end)) {
-                                u32 new_size;
-                                new_size = btrfs_file_extent_calc_inline_size(
-                                                   inline_limit - key.offset);
-                                inode_sub_bytes(inode, extent_end -
-                                                inline_limit);
-                                btrfs_set_file_extent_ram_bytes(leaf, extent,
-                                                        new_size);
-                                if (!compression && !encryption) {
-                                        btrfs_truncate_item(trans, root, path,
-                                                            new_size, 1);
-                                }
+                        extent_offset += end - key.offset;
+                        btrfs_set_file_extent_offset(leaf, fi, extent_offset);
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        extent_end - end);
+                        btrfs_mark_buffer_dirty(leaf);
+                        if (disk_bytenr > 0) {
+                                inode_sub_bytes(inode, end - key.offset);
+                                *hint_byte = disk_bytenr;
                         }
+                        break;
                 }
-                /* delete the entire extent */
-                if (!keep) {
-                        if (found_inline)
-                                inode_sub_bytes(inode, extent_end -
-                                                key.offset);
-                        ret = btrfs_del_item(trans, root, path);
-                        /* TODO update progress marker and return */
-                        BUG_ON(ret);
-                        extent = NULL;
-                        btrfs_release_path(root, path);
-                        /* the extent will be freed later */
-                }
-                if (bookend && found_inline && start <= key.offset) {
-                        u32 new_size;
-                        new_size = btrfs_file_extent_calc_inline_size(
-                                                   extent_end - end);
-                        inode_sub_bytes(inode, end - key.offset);
-                        btrfs_set_file_extent_ram_bytes(leaf, extent,
-                                                        new_size);
-                        if (!compression && !encryption)
-                                ret = btrfs_truncate_item(trans, root, path,
-                                                          new_size, 0);
-                        BUG_ON(ret);
-                }
-                /* create bookend, splitting the extent in two */
-                if (bookend && found_extent) {
-                        struct btrfs_key ins;
-                        ins.objectid = inode->i_ino;
-                        ins.offset = end;
-                        btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY);
 
-                        btrfs_release_path(root, path);
-                        path->leave_spinning = 1;
-                        ret = btrfs_insert_empty_item(trans, root, path, &ins,
-                                                      sizeof(*extent));
-                        BUG_ON(ret);
+                search_start = extent_end;
+                /*
+                 *       | ---- range to drop ----- |
+                 *  | -------- extent -------- |
+                 */
+                if (start > key.offset && end >= extent_end) {
+                        BUG_ON(del_nr > 0);
+                        BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);
 
-                        leaf = path->nodes[0];
-                        extent = btrfs_item_ptr(leaf, path->slots[0],
-                                                struct btrfs_file_extent_item);
-                        write_extent_buffer(leaf, &old,
-                                            (unsigned long)extent, sizeof(old));
-
-                        btrfs_set_file_extent_compression(leaf, extent,
-                                                          compression);
-                        btrfs_set_file_extent_encryption(leaf, extent,
-                                                         encryption);
-                        btrfs_set_file_extent_other_encoding(leaf, extent,
-                                                             other_encoding);
-                        btrfs_set_file_extent_offset(leaf, extent,
-                                    le64_to_cpu(old.offset) + end - key.offset);
-                        WARN_ON(le64_to_cpu(old.num_bytes) <
-                                (extent_end - end));
-                        btrfs_set_file_extent_num_bytes(leaf, extent,
-                                                        extent_end - end);
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        start - key.offset);
+                        btrfs_mark_buffer_dirty(leaf);
+                        if (disk_bytenr > 0) {
+                                inode_sub_bytes(inode, extent_end - start);
+                                *hint_byte = disk_bytenr;
+                        }
+                        if (end == extent_end)
+                                break;
 
-                        /*
-                         * set the ram bytes to the size of the full extent
-                         * before splitting.  This is a worst case flag,
-                         * but its the best we can do because we don't know
-                         * how splitting affects compression
-                         */
-                        btrfs_set_file_extent_ram_bytes(leaf, extent,
-                                                        ram_bytes);
-                        btrfs_set_file_extent_type(leaf, extent, found_type);
-
-                        btrfs_unlock_up_safe(path, 1);
-                        btrfs_mark_buffer_dirty(path->nodes[0]);
-                        btrfs_set_lock_blocking(path->nodes[0]);
-
-                        path->leave_spinning = 0;
-                        btrfs_release_path(root, path);
-                        if (disk_bytenr != 0)
-                                inode_add_bytes(inode, extent_end - end);
+                        path->slots[0]++;
+                        goto next_slot;
                 }
 
-                if (found_extent && !keep) {
-                        u64 old_disk_bytenr = le64_to_cpu(old.disk_bytenr);
+                /*
+                 *  | ---- range to drop ----- |
+                 *    | ------ extent ------ |
+                 */
+                if (start <= key.offset && end >= extent_end) {
+                        if (del_nr == 0) {
+                                del_slot = path->slots[0];
+                                del_nr = 1;
+                        } else {
+                                BUG_ON(del_slot + del_nr != path->slots[0]);
+                                del_nr++;
+                        }
 
-                        if (old_disk_bytenr != 0) {
+                        if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
                                 inode_sub_bytes(inode,
-                                                le64_to_cpu(old.num_bytes));
+                                                extent_end - key.offset);
+                                extent_end = ALIGN(extent_end,
+                                                   root->sectorsize);
+                        } else if (disk_bytenr > 0) {
                                 ret = btrfs_free_extent(trans, root,
-                                                old_disk_bytenr,
-                                                le64_to_cpu(old.disk_num_bytes),
-                                                0, root->root_key.objectid,
+                                                disk_bytenr, num_bytes, 0,
+                                                root->root_key.objectid,
                                                 key.objectid, key.offset -
-                                                le64_to_cpu(old.offset));
+                                                extent_offset);
                                 BUG_ON(ret);
-                                *hint_byte = old_disk_bytenr;
+                                inode_sub_bytes(inode,
+                                                extent_end - key.offset);
+                                *hint_byte = disk_bytenr;
                         }
-                }
 
-                if (search_start >= end) {
-                        ret = 0;
-                        goto out;
+                        if (end == extent_end)
+                                break;
+
+                        if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) {
+                                path->slots[0]++;
+                                goto next_slot;
+                        }
+
+                        ret = btrfs_del_items(trans, root, path, del_slot,
+                                              del_nr);
+                        BUG_ON(ret);
+
+                        del_nr = 0;
+                        del_slot = 0;
+
+                        btrfs_release_path(root, path);
+                        continue;
                 }
+
+                BUG_ON(1);
         }
-out:
-        btrfs_free_path(path);
-        if (locked_end > orig_locked_end) {
-                unlock_extent(&BTRFS_I(inode)->io_tree, orig_locked_end,
-                              locked_end - 1, GFP_NOFS);
+
+        if (del_nr > 0) {
+                ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
+                BUG_ON(ret);
         }
+
+        btrfs_free_path(path);
         return ret;
 }
 
 static int extent_mergeable(struct extent_buffer *leaf, int slot,
-                            u64 objectid, u64 bytenr, u64 *start, u64 *end)
+                            u64 objectid, u64 bytenr, u64 orig_offset,
+                            u64 *start, u64 *end)
 {
         struct btrfs_file_extent_item *fi;
         struct btrfs_key key;
@@ -598,6 +523,7 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot,
         fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
         if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG ||
             btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr ||
+            btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset ||
             btrfs_file_extent_compression(leaf, fi) ||
             btrfs_file_extent_encryption(leaf, fi) ||
             btrfs_file_extent_other_encoding(leaf, fi))
@@ -620,23 +546,24 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot,
  * two or three.
  */
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
-                              struct btrfs_root *root,
                               struct inode *inode, u64 start, u64 end)
 {
+        struct btrfs_root *root = BTRFS_I(inode)->root;
         struct extent_buffer *leaf;
         struct btrfs_path *path;
         struct btrfs_file_extent_item *fi;
         struct btrfs_key key;
+        struct btrfs_key new_key;
         u64 bytenr;
         u64 num_bytes;
         u64 extent_end;
         u64 orig_offset;
         u64 other_start;
         u64 other_end;
-        u64 split = start;
-        u64 locked_end = end;
-        int extent_type;
-        int split_end = 1;
+        u64 split;
+        int del_nr = 0;
+        int del_slot = 0;
+        int recow;
         int ret;
 
         btrfs_drop_extent_cache(inode, start, end - 1, 0);
@@ -644,12 +571,11 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
         path = btrfs_alloc_path();
         BUG_ON(!path);
 again:
+        recow = 0;
+        split = start;
         key.objectid = inode->i_ino;
         key.type = BTRFS_EXTENT_DATA_KEY;
-        if (split == start)
-                key.offset = split;
-        else
-                key.offset = split - 1;
+        key.offset = split;
 
         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
         if (ret > 0 && path->slots[0] > 0)
@@ -661,159 +587,158 @@ again:
                key.type != BTRFS_EXTENT_DATA_KEY);
         fi = btrfs_item_ptr(leaf, path->slots[0],
                             struct btrfs_file_extent_item);
-        extent_type = btrfs_file_extent_type(leaf, fi);
-        BUG_ON(extent_type != BTRFS_FILE_EXTENT_PREALLOC);
+        BUG_ON(btrfs_file_extent_type(leaf, fi) !=
+               BTRFS_FILE_EXTENT_PREALLOC);
         extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
         BUG_ON(key.offset > start || extent_end < end);
 
         bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
         num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
         orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi);
+        memcpy(&new_key, &key, sizeof(new_key));
 
-        if (key.offset == start)
-                split = end;
-
-        if (key.offset == start && extent_end == end) {
-                int del_nr = 0;
-                int del_slot = 0;
-                other_start = end;
-                other_end = 0;
-                if (extent_mergeable(leaf, path->slots[0] + 1, inode->i_ino,
-                                     bytenr, &other_start, &other_end)) {
-                        extent_end = other_end;
-                        del_slot = path->slots[0] + 1;
-                        del_nr++;
-                        ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
-                                                0, root->root_key.objectid,
-                                                inode->i_ino, orig_offset);
-                        BUG_ON(ret);
-                }
+        if (start == key.offset && end < extent_end) {
                 other_start = 0;
                 other_end = start;
-                if (extent_mergeable(leaf, path->slots[0] - 1, inode->i_ino,
-                                     bytenr, &other_start, &other_end)) {
-                        key.offset = other_start;
-                        del_slot = path->slots[0];
-                        del_nr++;
-                        ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
-                                                0, root->root_key.objectid,
-                                                inode->i_ino, orig_offset);
-                        BUG_ON(ret);
-                }
-                split_end = 0;
-                if (del_nr == 0) {
-                        btrfs_set_file_extent_type(leaf, fi,
-                                                   BTRFS_FILE_EXTENT_REG);
-                        goto done;
-                }
-
-                fi = btrfs_item_ptr(leaf, del_slot - 1,
-                                    struct btrfs_file_extent_item);
-                btrfs_set_file_extent_type(leaf, fi, BTRFS_FILE_EXTENT_REG);
-                btrfs_set_file_extent_num_bytes(leaf, fi,
-                                                extent_end - key.offset);
-                btrfs_mark_buffer_dirty(leaf);
-
-                ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
-                BUG_ON(ret);
-                goto release;
-        } else if (split == start) {
-                if (locked_end < extent_end) {
-                        ret = try_lock_extent(&BTRFS_I(inode)->io_tree,
-                                        locked_end, extent_end - 1, GFP_NOFS);
-                        if (!ret) {
-                                btrfs_release_path(root, path);
-                                lock_extent(&BTRFS_I(inode)->io_tree,
-                                        locked_end, extent_end - 1, GFP_NOFS);
-                                locked_end = extent_end;
-                                goto again;
-                        }
-                        locked_end = extent_end;
+                if (extent_mergeable(leaf, path->slots[0] - 1,
+                                     inode->i_ino, bytenr, orig_offset,
+                                     &other_start, &other_end)) {
+                        new_key.offset = end;
+                        btrfs_set_item_key_safe(trans, root, path, &new_key);
+                        fi = btrfs_item_ptr(leaf, path->slots[0],
+                                            struct btrfs_file_extent_item);
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        extent_end - end);
+                        btrfs_set_file_extent_offset(leaf, fi,
+                                                     end - orig_offset);
+                        fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
+                                            struct btrfs_file_extent_item);
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        end - other_start);
+                        btrfs_mark_buffer_dirty(leaf);
+                        goto out;
                 }
-                btrfs_set_file_extent_num_bytes(leaf, fi, split - key.offset);
-        } else  {
-                BUG_ON(key.offset != start);
-                key.offset = split;
-                btrfs_set_file_extent_offset(leaf, fi, key.offset -
-                                             orig_offset);
-                btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - split);
-                btrfs_set_item_key_safe(trans, root, path, &key);
-                extent_end = split;
         }
 
-        if (extent_end == end) {
-                split_end = 0;
-                extent_type = BTRFS_FILE_EXTENT_REG;
-        }
-        if (extent_end == end && split == start) {
+        if (start > key.offset && end == extent_end) {
                 other_start = end;
                 other_end = 0;
-                if (extent_mergeable(leaf, path->slots[0] + 1, inode->i_ino,
-                                     bytenr, &other_start, &other_end)) {
-                        path->slots[0]++;
+                if (extent_mergeable(leaf, path->slots[0] + 1,
+                                     inode->i_ino, bytenr, orig_offset,
+                                     &other_start, &other_end)) {
                         fi = btrfs_item_ptr(leaf, path->slots[0],
                                             struct btrfs_file_extent_item);
-                        key.offset = split;
-                        btrfs_set_item_key_safe(trans, root, path, &key);
-                        btrfs_set_file_extent_offset(leaf, fi, key.offset -
-                                                     orig_offset);
                         btrfs_set_file_extent_num_bytes(leaf, fi,
-                                                        other_end - split);
-                        goto done;
-                }
-        }
-        if (extent_end == end && split == end) {
-                other_start = 0;
-                other_end = start;
-                if (extent_mergeable(leaf, path->slots[0] - 1 , inode->i_ino,
-                                     bytenr, &other_start, &other_end)) {
-                        path->slots[0]--;
+                                                        start - key.offset);
+                        path->slots[0]++;
+                        new_key.offset = start;
+                        btrfs_set_item_key_safe(trans, root, path, &new_key);
+
                         fi = btrfs_item_ptr(leaf, path->slots[0],
                                             struct btrfs_file_extent_item);
-                        btrfs_set_file_extent_num_bytes(leaf, fi, extent_end -
-                                                        other_start);
-                        goto done;
+                        btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                        other_end - start);
+                        btrfs_set_file_extent_offset(leaf, fi,
+                                                     start - orig_offset);
+                        btrfs_mark_buffer_dirty(leaf);
+                        goto out;
                 }
         }
 
-        btrfs_mark_buffer_dirty(leaf);
+        while (start > key.offset || end < extent_end) {
+                if (key.offset == start)
+                        split = end;
 
-        ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0,
-                                   root->root_key.objectid,
-                                   inode->i_ino, orig_offset);
-        BUG_ON(ret);
-        btrfs_release_path(root, path);
+                new_key.offset = split;
+                ret = btrfs_duplicate_item(trans, root, path, &new_key);
+                if (ret == -EAGAIN) {
+                        btrfs_release_path(root, path);
+                        goto again;
+                }
+                BUG_ON(ret < 0);
 
-        key.offset = start;
-        ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(*fi));
-        BUG_ON(ret);
+                leaf = path->nodes[0];
+                fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
+                                    struct btrfs_file_extent_item);
+                btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                split - key.offset);
 
-        leaf = path->nodes[0];
-        fi = btrfs_item_ptr(leaf, path->slots[0],
-                            struct btrfs_file_extent_item);
-        btrfs_set_file_extent_generation(leaf, fi, trans->transid);
-        btrfs_set_file_extent_type(leaf, fi, extent_type);
-        btrfs_set_file_extent_disk_bytenr(leaf, fi, bytenr);
-        btrfs_set_file_extent_disk_num_bytes(leaf, fi, num_bytes);
-        btrfs_set_file_extent_offset(leaf, fi, key.offset - orig_offset);
-        btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - key.offset);
-        btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
-        btrfs_set_file_extent_compression(leaf, fi, 0);
-        btrfs_set_file_extent_encryption(leaf, fi, 0);
-        btrfs_set_file_extent_other_encoding(leaf, fi, 0);
-done:
-        btrfs_mark_buffer_dirty(leaf);
-
-release:
-        btrfs_release_path(root, path);
-        if (split_end && split == start) {
-                split = end;
-                goto again;
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                                    struct btrfs_file_extent_item);
+
+                btrfs_set_file_extent_offset(leaf, fi, split - orig_offset);
+                btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                extent_end - split);
+                btrfs_mark_buffer_dirty(leaf);
+
+                ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0,
+                                           root->root_key.objectid,
+                                           inode->i_ino, orig_offset);
+                BUG_ON(ret);
+
+                if (split == start) {
+                        key.offset = start;
+                } else {
+                        BUG_ON(start != key.offset);
+                        path->slots[0]--;
+                        extent_end = end;
+                }
+                recow = 1;
         }
-        if (locked_end > end) {
-                unlock_extent(&BTRFS_I(inode)->io_tree, end, locked_end - 1,
-                              GFP_NOFS);
+
+        other_start = end;
+        other_end = 0;
+        if (extent_mergeable(leaf, path->slots[0] + 1,
+                             inode->i_ino, bytenr, orig_offset,
+                             &other_start, &other_end)) {
+                if (recow) {
+                        btrfs_release_path(root, path);
+                        goto again;
+                }
+                extent_end = other_end;
+                del_slot = path->slots[0] + 1;
+                del_nr++;
+                ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
+                                        0, root->root_key.objectid,
+                                        inode->i_ino, orig_offset);
+                BUG_ON(ret);
         }
+        other_start = 0;
+        other_end = start;
+        if (extent_mergeable(leaf, path->slots[0] - 1,
+                             inode->i_ino, bytenr, orig_offset,
+                             &other_start, &other_end)) {
+                if (recow) {
+                        btrfs_release_path(root, path);
+                        goto again;
+                }
+                key.offset = other_start;
+                del_slot = path->slots[0];
+                del_nr++;
+                ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
+                                        0, root->root_key.objectid,
+                                        inode->i_ino, orig_offset);
+                BUG_ON(ret);
+        }
+        if (del_nr == 0) {
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                           struct btrfs_file_extent_item);
+                btrfs_set_file_extent_type(leaf, fi,
+                                           BTRFS_FILE_EXTENT_REG);
+                btrfs_mark_buffer_dirty(leaf);
+        } else {
+                fi = btrfs_item_ptr(leaf, del_slot - 1,
+                           struct btrfs_file_extent_item);
+                btrfs_set_file_extent_type(leaf, fi,
+                                           BTRFS_FILE_EXTENT_REG);
+                btrfs_set_file_extent_num_bytes(leaf, fi,
+                                                extent_end - key.offset);
+                btrfs_mark_buffer_dirty(leaf);
+
+                ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
+                BUG_ON(ret);
+        }
+out:
         btrfs_free_path(path);
         return 0;
 }
@@ -909,7 +834,7 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
         unsigned long last_index;
         int will_write;
 
-        will_write = ((file->f_flags & O_SYNC) || IS_SYNC(inode) ||
+        will_write = ((file->f_flags & O_DSYNC) || IS_SYNC(inode) ||
                       (file->f_flags & O_DIRECT));
 
         nrptrs = min((count + PAGE_CACHE_SIZE - 1) / PAGE_CACHE_SIZE,
@@ -1076,7 +1001,7 @@ out_nolock:
                 if (err)
                         num_written = err;
 
-                if ((file->f_flags & O_SYNC) || IS_SYNC(inode)) {
+                if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) {
                         trans = btrfs_start_transaction(root, 1);
                         ret = btrfs_log_dentry_safe(trans, root,
                                                     file->f_dentry);
@@ -1086,8 +1011,10 @@ out_nolock:
                                         btrfs_end_transaction(trans, root);
                                 else
                                         btrfs_commit_transaction(trans, root);
-                        } else {
+                        } else if (ret != BTRFS_NO_LOG_SYNC) {
                                 btrfs_commit_transaction(trans, root);
+                        } else {
+                                btrfs_end_transaction(trans, root);
                         }
                 }
                 if (file->f_flags & O_DIRECT) {
@@ -1137,6 +1064,13 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
         int ret = 0;
         struct btrfs_trans_handle *trans;
 
+
+        /* we wait first, since the writeback may change the inode */
+        root->log_batch++;
+        /* the VFS called filemap_fdatawrite for us */
+        btrfs_wait_ordered_range(inode, 0, (u64)-1);
+        root->log_batch++;
+
         /*
          * check the transaction that last modified this inode
          * and see if its already been committed
@@ -1144,6 +1078,11 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
         if (!BTRFS_I(inode)->last_trans)
                 goto out;
 
+        /*
+         * if the last transaction that changed this file was before
+         * the current transaction, we can bail out now without any
+         * syncing
+         */
         mutex_lock(&root->fs_info->trans_mutex);
         if (BTRFS_I(inode)->last_trans <=
             root->fs_info->last_trans_committed) {
@@ -1153,13 +1092,6 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
         }
         mutex_unlock(&root->fs_info->trans_mutex);
 
-        root->log_batch++;
-        filemap_fdatawrite(inode->i_mapping);
-        btrfs_wait_ordered_range(inode, 0, (u64)-1);
-        root->log_batch++;
-
-        if (datasync && !(inode->i_state & I_DIRTY_PAGES))
-                goto out;
         /*
          * ok we haven't committed the transaction yet, lets do a commit
          */
@@ -1188,18 +1120,22 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
          */
         mutex_unlock(&dentry->d_inode->i_mutex);
 
-        if (ret > 0) {
-                ret = btrfs_commit_transaction(trans, root);
-        } else {
-                ret = btrfs_sync_log(trans, root);
-                if (ret == 0)
-                        ret = btrfs_end_transaction(trans, root);
-                else
+        if (ret != BTRFS_NO_LOG_SYNC) {
+                if (ret > 0) {
                         ret = btrfs_commit_transaction(trans, root);
+                } else {
+                        ret = btrfs_sync_log(trans, root);
+                        if (ret == 0)
+                                ret = btrfs_end_transaction(trans, root);
+                        else
+                                ret = btrfs_commit_transaction(trans, root);
+                }
+        } else {
+                ret = btrfs_end_transaction(trans, root);
         }
         mutex_lock(&dentry->d_inode->i_mutex);
 out:
-        return ret > 0 ? EIO : ret;
+        return ret > 0 ? -EIO : ret;
 }
 
 static const struct vm_operations_struct btrfs_file_vm_ops = {
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 5c2caad76212..cb2849f03251 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -1296,7 +1296,7 @@ again:
                         window_start = entry->offset;
                         window_free = entry->bytes;
                         last = entry;
-                        max_extent = 0;
+                        max_extent = entry->bytes;
                 } else {
                         last = next;
                         window_free += next->bytes;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 9e138b793dc7..4deb280f8969 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -88,13 +88,14 @@ static noinline int cow_file_range(struct inode *inode,
                                    u64 start, u64 end, int *page_started,
                                    unsigned long *nr_written, int unlock);
 
-static int btrfs_init_inode_security(struct inode *inode,  struct inode *dir)
+static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
+                                     struct inode *inode,  struct inode *dir)
 {
         int err;
 
-        err = btrfs_init_acl(inode, dir);
+        err = btrfs_init_acl(trans, inode, dir);
         if (!err)
-                err = btrfs_xattr_security_init(inode, dir);
+                err = btrfs_xattr_security_init(trans, inode, dir);
         return err;
 }
 
@@ -188,8 +189,18 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans,
         btrfs_mark_buffer_dirty(leaf);
         btrfs_free_path(path);
 
+        /*
+         * we're an inline extent, so nobody can
+         * extend the file past i_size without locking
+         * a page we already have locked.
+         *
+         * We must do any isize and inode updates
+         * before we unlock the pages.  Otherwise we
+         * could end up racing with unlink.
+         */
         BTRFS_I(inode)->disk_i_size = inode->i_size;
         btrfs_update_inode(trans, root, inode);
+
         return 0;
 fail:
         btrfs_free_path(path);
@@ -230,8 +241,7 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
                 return 1;
         }
 
-        ret = btrfs_drop_extents(trans, root, inode, start,
-                                 aligned_end, aligned_end, start,
+        ret = btrfs_drop_extents(trans, inode, start, aligned_end,
                                  &hint_byte, 1);
         BUG_ON(ret);
 
@@ -416,7 +426,6 @@ again:
                                                     start, end,
                                                     total_compressed, pages);
                 }
-                btrfs_end_transaction(trans, root);
                 if (ret == 0) {
                         /*
                          * inline extent creation worked, we don't need
@@ -430,9 +439,11 @@ again:
                              EXTENT_CLEAR_DELALLOC |
                              EXTENT_CLEAR_ACCOUNTING |
                              EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK);
-                        ret = 0;
+
+                        btrfs_end_transaction(trans, root);
                         goto free_pages_out;
                 }
+                btrfs_end_transaction(trans, root);
         }
 
         if (will_compress) {
@@ -472,7 +483,8 @@ again:
                 nr_pages_ret = 0;
 
                 /* flag the file so we don't compress in the future */
-                BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
+                if (!btrfs_test_opt(root, FORCE_COMPRESS))
+                        BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
         }
         if (will_compress) {
                 *num_added += 1;
@@ -538,12 +550,11 @@ static noinline int submit_compressed_extents(struct inode *inode,
         struct btrfs_root *root = BTRFS_I(inode)->root;
         struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
         struct extent_io_tree *io_tree;
-        int ret;
+        int ret = 0;
 
         if (list_empty(&async_cow->extents))
                 return 0;
 
-        trans = btrfs_join_transaction(root, 1);
 
         while (!list_empty(&async_cow->extents)) {
                 async_extent = list_entry(async_cow->extents.next,
@@ -552,6 +563,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 
                 io_tree = &BTRFS_I(inode)->io_tree;
 
+retry:
                 /* did the compression code fall back to uncompressed IO? */
                 if (!async_extent->pages) {
                         int page_started = 0;
@@ -562,11 +574,11 @@ static noinline int submit_compressed_extents(struct inode *inode,
                                     async_extent->ram_size - 1, GFP_NOFS);
 
                         /* allocate blocks */
-                        cow_file_range(inode, async_cow->locked_page,
-                                       async_extent->start,
-                                       async_extent->start +
-                                       async_extent->ram_size - 1,
-                                       &page_started, &nr_written, 0);
+                        ret = cow_file_range(inode, async_cow->locked_page,
+                                             async_extent->start,
+                                             async_extent->start +
+                                             async_extent->ram_size - 1,
+                                             &page_started, &nr_written, 0);
 
                         /*
                          * if page_started, cow_file_range inserted an
@@ -574,7 +586,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
                          * and IO for us.  Otherwise, we need to submit
                          * all those pages down to the drive.
                          */
-                        if (!page_started)
+                        if (!page_started && !ret)
                                 extent_write_locked_range(io_tree,
                                                   inode, async_extent->start,
                                                   async_extent->start +
@@ -589,6 +601,30 @@ static noinline int submit_compressed_extents(struct inode *inode,
                 lock_extent(io_tree, async_extent->start,
                             async_extent->start + async_extent->ram_size - 1,
                             GFP_NOFS);
+
+                trans = btrfs_join_transaction(root, 1);
+                ret = btrfs_reserve_extent(trans, root,
+                                           async_extent->compressed_size,
+                                           async_extent->compressed_size,
+                                           0, alloc_hint,
+                                           (u64)-1, &ins, 1);
+                btrfs_end_transaction(trans, root);
+
+                if (ret) {
+                        int i;
+                        for (i = 0; i < async_extent->nr_pages; i++) {
+                                WARN_ON(async_extent->pages[i]->mapping);
+                                page_cache_release(async_extent->pages[i]);
+                        }
+                        kfree(async_extent->pages);
+                        async_extent->nr_pages = 0;
+                        async_extent->pages = NULL;
+                        unlock_extent(io_tree, async_extent->start,
+                                      async_extent->start +
+                                      async_extent->ram_size - 1, GFP_NOFS);
+                        goto retry;
+                }
+
                 /*
                  * here we're doing allocation and writeback of the
                  * compressed pages
@@ -597,12 +633,6 @@ static noinline int submit_compressed_extents(struct inode *inode,
                                         async_extent->start +
                                         async_extent->ram_size - 1, 0);
 
-                ret = btrfs_reserve_extent(trans, root,
-                                           async_extent->compressed_size,
-                                           async_extent->compressed_size,
-                                           0, alloc_hint,
-                                           (u64)-1, &ins, 1);
-                BUG_ON(ret);
                 em = alloc_extent_map(GFP_NOFS);
                 em->start = async_extent->start;
                 em->len = async_extent->ram_size;
@@ -634,8 +664,6 @@ static noinline int submit_compressed_extents(struct inode *inode,
                                                BTRFS_ORDERED_COMPRESSED);
                 BUG_ON(ret);
 
-                btrfs_end_transaction(trans, root);
-
                 /*
                  * clear dirty, set writeback and unlock the pages.
                  */
@@ -657,13 +685,11 @@ static noinline int submit_compressed_extents(struct inode *inode,
                                     async_extent->nr_pages);
 
                 BUG_ON(ret);
-                trans = btrfs_join_transaction(root, 1);
                 alloc_hint = ins.objectid + ins.offset;
                 kfree(async_extent);
                 cond_resched();
         }
 
-        btrfs_end_transaction(trans, root);
         return 0;
 }
 
@@ -727,6 +753,7 @@ static noinline int cow_file_range(struct inode *inode,
                                      EXTENT_CLEAR_DIRTY |
                                      EXTENT_SET_WRITEBACK |
                                      EXTENT_END_WRITEBACK);
+
                         *nr_written = *nr_written +
                              (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE;
                         *page_started = 1;
@@ -743,8 +770,22 @@ static noinline int cow_file_range(struct inode *inode,
         em = search_extent_mapping(&BTRFS_I(inode)->extent_tree,
                                    start, num_bytes);
         if (em) {
-                alloc_hint = em->block_start;
-                free_extent_map(em);
+                /*
+                 * if block start isn't an actual block number then find the
+                 * first block in this inode and use that as a hint.  If that
+                 * block is also bogus then just don't worry about it.
+                 */
+                if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
+                        free_extent_map(em);
+                        em = search_extent_mapping(em_tree, 0, 0);
+                        if (em && em->block_start < EXTENT_MAP_LAST_BYTE)
+                                alloc_hint = em->block_start;
+                        if (em)
+                                free_extent_map(em);
+                } else {
+                        alloc_hint = em->block_start;
+                        free_extent_map(em);
+                }
         }
         read_unlock(&BTRFS_I(inode)->extent_tree.lock);
         btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);
@@ -1567,7 +1608,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
                                        struct inode *inode, u64 file_pos,
                                        u64 disk_bytenr, u64 disk_num_bytes,
                                        u64 num_bytes, u64 ram_bytes,
-                                       u64 locked_end,
                                        u8 compression, u8 encryption,
                                        u16 other_encoding, int extent_type)
 {
@@ -1593,9 +1633,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
          * the caller is expected to unpin it and allow it to be merged
          * with the others.
          */
-        ret = btrfs_drop_extents(trans, root, inode, file_pos,
-                                 file_pos + num_bytes, locked_end,
-                                 file_pos, &hint, 0);
+        ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes,
+                                 &hint, 0);
         BUG_ON(ret);
 
         ins.objectid = inode->i_ino;
@@ -1642,24 +1681,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
  * before we start the transaction.  It limits the amount of btree
  * reads required while inside the transaction.
  */
-static noinline void reada_csum(struct btrfs_root *root,
-                                struct btrfs_path *path,
-                                struct btrfs_ordered_extent *ordered_extent)
-{
-        struct btrfs_ordered_sum *sum;
-        u64 bytenr;
-
-        sum = list_entry(ordered_extent->list.next, struct btrfs_ordered_sum,
-                         list);
-        bytenr = sum->sums[0].bytenr;
-
-        /*
-         * we don't care about the results, the point of this search is
-         * just to get the btree leaves into ram
-         */
-        btrfs_lookup_csum(NULL, root->fs_info->csum_root, path, bytenr, 0);
-}
-
 /* as ordered data IO finishes, this gets called so we can finish
  * an ordered extent if the range of bytes in the file it covers are
  * fully written.
@@ -1670,7 +1691,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
         struct btrfs_trans_handle *trans;
         struct btrfs_ordered_extent *ordered_extent = NULL;
         struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
-        struct btrfs_path *path;
         int compressed = 0;
         int ret;
 
@@ -1678,46 +1698,32 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
         if (!ret)
                 return 0;
 
-        /*
-         * before we join the transaction, try to do some of our IO.
-         * This will limit the amount of IO that we have to do with
-         * the transaction running.  We're unlikely to need to do any
-         * IO if the file extents are new, the disk_i_size checks
-         * covers the most common case.
-         */
-        if (start < BTRFS_I(inode)->disk_i_size) {
-                path = btrfs_alloc_path();
-                if (path) {
-                        ret = btrfs_lookup_file_extent(NULL, root, path,
-                                                       inode->i_ino,
-                                                       start, 0);
-                        ordered_extent = btrfs_lookup_ordered_extent(inode,
-                                                                     start);
-                        if (!list_empty(&ordered_extent->list)) {
-                                btrfs_release_path(root, path);
-                                reada_csum(root, path, ordered_extent);
-                        }
-                        btrfs_free_path(path);
+        ordered_extent = btrfs_lookup_ordered_extent(inode, start);
+        BUG_ON(!ordered_extent);
+
+        if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
+                BUG_ON(!list_empty(&ordered_extent->list));
+                ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
+                if (!ret) {
+                        trans = btrfs_join_transaction(root, 1);
+                        ret = btrfs_update_inode(trans, root, inode);
+                        BUG_ON(ret);
+                        btrfs_end_transaction(trans, root);
                 }
+                goto out;
         }
 
-        trans = btrfs_join_transaction(root, 1);
-
-        if (!ordered_extent)
-                ordered_extent = btrfs_lookup_ordered_extent(inode, start);
-        BUG_ON(!ordered_extent);
-        if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags))
-                goto nocow;
-
         lock_extent(io_tree, ordered_extent->file_offset,
                     ordered_extent->file_offset + ordered_extent->len - 1,
                     GFP_NOFS);
 
+        trans = btrfs_join_transaction(root, 1);
+
         if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
                 compressed = 1;
         if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
                 BUG_ON(compressed);
-                ret = btrfs_mark_extent_written(trans, root, inode,
+                ret = btrfs_mark_extent_written(trans, inode,
                                                 ordered_extent->file_offset,
                                                 ordered_extent->file_offset +
                                                 ordered_extent->len);
@@ -1729,8 +1735,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
                                                 ordered_extent->disk_len,
                                                 ordered_extent->len,
                                                 ordered_extent->len,
-                                                ordered_extent->file_offset +
-                                                ordered_extent->len,
                                                 compressed, 0, 0,
                                                 BTRFS_FILE_EXTENT_REG);
                 unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
@@ -1741,22 +1745,20 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
         unlock_extent(io_tree, ordered_extent->file_offset,
                     ordered_extent->file_offset + ordered_extent->len - 1,
                     GFP_NOFS);
-nocow:
         add_pending_csums(trans, inode, ordered_extent->file_offset,
                           &ordered_extent->list);
 
-        mutex_lock(&BTRFS_I(inode)->extent_mutex);
-        btrfs_ordered_update_i_size(inode, ordered_extent);
-        btrfs_update_inode(trans, root, inode);
-        btrfs_remove_ordered_extent(inode, ordered_extent);
-        mutex_unlock(&BTRFS_I(inode)->extent_mutex);
-
+        /* this also removes the ordered extent from the tree */
+        btrfs_ordered_update_i_size(inode, 0, ordered_extent);
+        ret = btrfs_update_inode(trans, root, inode);
+        BUG_ON(ret);
+        btrfs_end_transaction(trans, root);
+out:
         /* once for us */
         btrfs_put_ordered_extent(ordered_extent);
         /* once for the tree */
         btrfs_put_ordered_extent(ordered_extent);
 
-        btrfs_end_transaction(trans, root);
         return 0;
 }
 
@@ -1979,6 +1981,54 @@ zeroit:
         return -EIO;
 }
 
+struct delayed_iput {
+        struct list_head list;
+        struct inode *inode;
+};
+
+void btrfs_add_delayed_iput(struct inode *inode)
+{
+        struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
+        struct delayed_iput *delayed;
+
+        if (atomic_add_unless(&inode->i_count, -1, 1))
+                return;
+
+        delayed = kmalloc(sizeof(*delayed), GFP_NOFS | __GFP_NOFAIL);
+        delayed->inode = inode;
+
+        spin_lock(&fs_info->delayed_iput_lock);
+        list_add_tail(&delayed->list, &fs_info->delayed_iputs);
+        spin_unlock(&fs_info->delayed_iput_lock);
+}
+
+void btrfs_run_delayed_iputs(struct btrfs_root *root)
+{
+        LIST_HEAD(list);
+        struct btrfs_fs_info *fs_info = root->fs_info;
+        struct delayed_iput *delayed;
+        int empty;
+
+        spin_lock(&fs_info->delayed_iput_lock);
+        empty = list_empty(&fs_info->delayed_iputs);
+        spin_unlock(&fs_info->delayed_iput_lock);
+        if (empty)
+                return;
+
+        down_read(&root->fs_info->cleanup_work_sem);
+        spin_lock(&fs_info->delayed_iput_lock);
+        list_splice_init(&fs_info->delayed_iputs, &list);
+        spin_unlock(&fs_info->delayed_iput_lock);
+
+        while (!list_empty(&list)) {
+                delayed = list_entry(list.next, struct delayed_iput, list);
+                list_del(&delayed->list);
+                iput(delayed->inode);
+                kfree(delayed);
+        }
+        up_read(&root->fs_info->cleanup_work_sem);
+}
+
 /*
  * This creates an orphan entry for the given inode in case something goes
  * wrong in the middle of an unlink/truncate.
@@ -2051,16 +2101,17 @@ void btrfs_orphan_cleanup(struct btrfs_root *root)
         struct inode *inode;
         int ret = 0, nr_unlink = 0, nr_truncate = 0;
 
-        path = btrfs_alloc_path();
-        if (!path)
+        if (!xchg(&root->clean_orphans, 0))
                 return;
+
+        path = btrfs_alloc_path();
+        BUG_ON(!path);
         path->reada = -1;
 
         key.objectid = BTRFS_ORPHAN_OBJECTID;
         btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
         key.offset = (u64)-1;
 
-
         while (1) {
                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
                 if (ret < 0) {
@@ -2474,7 +2525,19 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 
         root = BTRFS_I(dir)->root;
 
+        /*
+         * 5 items for unlink inode
+         * 1 for orphan
+         */
+        ret = btrfs_reserve_metadata_space(root, 6);
+        if (ret)
+                return ret;
+
         trans = btrfs_start_transaction(root, 1);
+        if (IS_ERR(trans)) {
+                btrfs_unreserve_metadata_space(root, 6);
+                return PTR_ERR(trans);
+        }
 
         btrfs_set_trans_block_group(trans, dir);
 
@@ -2489,6 +2552,7 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
         nr = trans->blocks_used;
 
         btrfs_end_transaction_throttle(trans, root);
+        btrfs_unreserve_metadata_space(root, 6);
         btrfs_btree_balance_dirty(root, nr);
         return ret;
 }
@@ -2569,7 +2633,16 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
             inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
                 return -ENOTEMPTY;
 
+        ret = btrfs_reserve_metadata_space(root, 5);
+        if (ret)
+                return ret;
+
         trans = btrfs_start_transaction(root, 1);
+        if (IS_ERR(trans)) {
+                btrfs_unreserve_metadata_space(root, 5);
+                return PTR_ERR(trans);
+        }
+
         btrfs_set_trans_block_group(trans, dir);
 
         if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
@@ -2592,6 +2665,7 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 out:
         nr = trans->blocks_used;
         ret = btrfs_end_transaction_throttle(trans, root);
+        btrfs_unreserve_metadata_space(root, 5);
         btrfs_btree_balance_dirty(root, nr);
 
         if (ret && !err)
@@ -2782,37 +2856,40 @@ out:
  * min_type is the minimum key type to truncate down to.  If set to 0, this
  * will kill all the items on this inode, including the INODE_ITEM_KEY.
  */
-noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
-                                        struct btrfs_root *root,
-                                        struct inode *inode,
-                                        u64 new_size, u32 min_type)
+int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
+                               struct btrfs_root *root,
+                               struct inode *inode,
+                               u64 new_size, u32 min_type)
 {
-        int ret;
         struct btrfs_path *path;
-        struct btrfs_key key;
-        struct btrfs_key found_key;
-        u32 found_type = (u8)-1;
         struct extent_buffer *leaf;
         struct btrfs_file_extent_item *fi;
+        struct btrfs_key key;
+        struct btrfs_key found_key;
         u64 extent_start = 0;
         u64 extent_num_bytes = 0;
         u64 extent_offset = 0;
         u64 item_end = 0;
+        u64 mask = root->sectorsize - 1;
+        u32 found_type = (u8)-1;
         int found_extent;
         int del_item;
         int pending_del_nr = 0;
         int pending_del_slot = 0;
         int extent_type = -1;
         int encoding;
-        u64 mask = root->sectorsize - 1;
+        int ret;
+        int err = 0;
+
+        BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY);
 
         if (root->ref_cows)
                 btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0);
+
         path = btrfs_alloc_path();
         BUG_ON(!path);
         path->reada = -1;
 
-        /* FIXME, add redo link to tree so we don't leak on crash */
         key.objectid = inode->i_ino;
         key.offset = (u64)-1;
         key.type = (u8)-1;
@@ -2820,17 +2897,17 @@ noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
 search_again:
         path->leave_spinning = 1;
         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
-        if (ret < 0)
-                goto error;
+        if (ret < 0) {
+                err = ret;
+                goto out;
+        }
 
         if (ret > 0) {
                 /* there are no items in the tree for us to truncate, we're
                  * done
                  */
-                if (path->slots[0] == 0) {
-                        ret = 0;
-                        goto error;
-                }
+                if (path->slots[0] == 0)
+                        goto out;
                 path->slots[0]--;
         }
 
@@ -2865,28 +2942,17 @@ search_again:
                         }
                         item_end--;
                 }
-                if (item_end < new_size) {
-                        if (found_type == BTRFS_DIR_ITEM_KEY)
-                                found_type = BTRFS_INODE_ITEM_KEY;
-                        else if (found_type == BTRFS_EXTENT_ITEM_KEY)
-                                found_type = BTRFS_EXTENT_DATA_KEY;
-                        else if (found_type == BTRFS_EXTENT_DATA_KEY)
-                                found_type = BTRFS_XATTR_ITEM_KEY;
-                        else if (found_type == BTRFS_XATTR_ITEM_KEY)
-                                found_type = BTRFS_INODE_REF_KEY;
-                        else if (found_type)
-                                found_type--;
-                        else
+                if (found_type > min_type) {
+                        del_item = 1;
+                } else {
+                        if (item_end < new_size)
                                 break;
-                        btrfs_set_key_type(&key, found_type);
-                        goto next;
+                        if (found_key.offset >= new_size)
+                                del_item = 1;
+                        else
+                                del_item = 0;
                 }
-                if (found_key.offset >= new_size)
-                        del_item = 1;
-                else
-                        del_item = 0;
                 found_extent = 0;
-
                 /* FIXME, shrink the extent if the ref count is only 1 */
                 if (found_type != BTRFS_EXTENT_DATA_KEY)
                         goto delete;
@@ -2973,42 +3039,36 @@ delete:
                                                 inode->i_ino, extent_offset);
                         BUG_ON(ret);
                 }
-next:
-                if (path->slots[0] == 0) {
-                        if (pending_del_nr)
-                                goto del_pending;
-                        btrfs_release_path(root, path);
-                        if (found_type == BTRFS_INODE_ITEM_KEY)
-                                break;
-                        goto search_again;
-                }
 
-                path->slots[0]--;
-                if (pending_del_nr &&
-                    path->slots[0] + 1 != pending_del_slot) {
-                        struct btrfs_key debug;
-del_pending:
-                        btrfs_item_key_to_cpu(path->nodes[0], &debug,
-                                              pending_del_slot);
-                        ret = btrfs_del_items(trans, root, path,
-                                              pending_del_slot,
-                                              pending_del_nr);
-                        BUG_ON(ret);
-                        pending_del_nr = 0;
+                if (found_type == BTRFS_INODE_ITEM_KEY)
+                        break;
+
+                if (path->slots[0] == 0 ||
+                    path->slots[0] != pending_del_slot) {
+                        if (root->ref_cows) {
+                                err = -EAGAIN;
+                                goto out;
+                        }
+                        if (pending_del_nr) {
+                                ret = btrfs_del_items(trans, root, path,
+                                                pending_del_slot,
+                                                pending_del_nr);
+                                BUG_ON(ret);
+                                pending_del_nr = 0;
+                        }
                         btrfs_release_path(root, path);
-                        if (found_type == BTRFS_INODE_ITEM_KEY)
-                                break;
                         goto search_again;
+                } else {
+                        path->slots[0]--;
                 }
         }
-        ret = 0;
-error:
+out:
         if (pending_del_nr) {
                 ret = btrfs_del_items(trans, root, path, pending_del_slot,
                                       pending_del_nr);
         }
         btrfs_free_path(path);
-        return ret;
+        return err;
 }
 
 /*
@@ -3032,12 +3092,22 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
 
         if ((offset & (blocksize - 1)) == 0)
                 goto out;
+        ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE);
+        if (ret)
+                goto out;
+
+        ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1);
+        if (ret)
+                goto out;
 
         ret = -ENOMEM;
 again:
         page = grab_cache_page(mapping, index);
-        if (!page)
+        if (!page) {
+                btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+                btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
                 goto out;
+        }
 
         page_start = page_offset(page);
         page_end = page_start + PAGE_CACHE_SIZE - 1;
@@ -3070,6 +3140,10 @@ again:
                 goto again;
         }
 
+        clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
+                          EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+                          GFP_NOFS);
+
         ret = btrfs_set_extent_delalloc(inode, page_start, page_end);
         if (ret) {
                 unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
@@ -3088,6 +3162,9 @@ again:
         unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 
 out_unlock:
+        if (ret)
+                btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+        btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
         unlock_page(page);
         page_cache_release(page);
 out:
@@ -3111,8 +3188,6 @@ int btrfs_cont_expand(struct inode *inode, loff_t size)
         if (size <= hole_start)
                 return 0;
 
-        btrfs_truncate_page(inode->i_mapping, inode->i_size);
-
         while (1) {
                 struct btrfs_ordered_extent *ordered;
                 btrfs_wait_ordered_range(inode, hole_start,
@@ -3125,9 +3200,6 @@ int btrfs_cont_expand(struct inode *inode, loff_t size)
                 btrfs_put_ordered_extent(ordered);
         }
 
-        trans = btrfs_start_transaction(root, 1);
-        btrfs_set_trans_block_group(trans, inode);
-
         cur_offset = hole_start;
         while (1) {
                 em = btrfs_get_extent(inode, NULL, 0, cur_offset,
@@ -3135,40 +3207,120 @@ int btrfs_cont_expand(struct inode *inode, loff_t size)
                 BUG_ON(IS_ERR(em) || !em);
                 last_byte = min(extent_map_end(em), block_end);
                 last_byte = (last_byte + mask) & ~mask;
-                if (test_bit(EXTENT_FLAG_VACANCY, &em->flags)) {
+                if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
                         u64 hint_byte = 0;
                         hole_size = last_byte - cur_offset;
-                        err = btrfs_drop_extents(trans, root, inode,
-                                                 cur_offset,
-                                                 cur_offset + hole_size,
-                                                 block_end,
-                                                 cur_offset, &hint_byte, 1);
-                        if (err)
-                                break;
 
-                        err = btrfs_reserve_metadata_space(root, 1);
+                        err = btrfs_reserve_metadata_space(root, 2);
                         if (err)
                                 break;
 
+                        trans = btrfs_start_transaction(root, 1);
+                        btrfs_set_trans_block_group(trans, inode);
+
+                        err = btrfs_drop_extents(trans, inode, cur_offset,
+                                                 cur_offset + hole_size,
+                                                 &hint_byte, 1);
+                        BUG_ON(err);
+
                         err = btrfs_insert_file_extent(trans, root,
                                         inode->i_ino, cur_offset, 0,
                                         0, hole_size, 0, hole_size,
                                         0, 0, 0);
+                        BUG_ON(err);
+
                         btrfs_drop_extent_cache(inode, hole_start,
                                         last_byte - 1, 0);
-                        btrfs_unreserve_metadata_space(root, 1);
+
+                        btrfs_end_transaction(trans, root);
+                        btrfs_unreserve_metadata_space(root, 2);
                 }
                 free_extent_map(em);
                 cur_offset = last_byte;
-                if (err || cur_offset >= block_end)
+                if (cur_offset >= block_end)
                         break;
         }
 
-        btrfs_end_transaction(trans, root);
         unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
         return err;
 }
 
+static int btrfs_setattr_size(struct inode *inode, struct iattr *attr)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        struct btrfs_trans_handle *trans;
+        unsigned long nr;
+        int ret;
+
+        if (attr->ia_size == inode->i_size)
+                return 0;
+
+        if (attr->ia_size > inode->i_size) {
+                unsigned long limit;
+                limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
+                if (attr->ia_size > inode->i_sb->s_maxbytes)
+                        return -EFBIG;
+                if (limit != RLIM_INFINITY && attr->ia_size > limit) {
+                        send_sig(SIGXFSZ, current, 0);
+                        return -EFBIG;
+                }
+        }
+
+        ret = btrfs_reserve_metadata_space(root, 1);
+        if (ret)
+                return ret;
+
+        trans = btrfs_start_transaction(root, 1);
+        btrfs_set_trans_block_group(trans, inode);
+
+        ret = btrfs_orphan_add(trans, inode);
+        BUG_ON(ret);
+
+        nr = trans->blocks_used;
+        btrfs_end_transaction(trans, root);
+        btrfs_unreserve_metadata_space(root, 1);
+        btrfs_btree_balance_dirty(root, nr);
+
+        if (attr->ia_size > inode->i_size) {
+                ret = btrfs_cont_expand(inode, attr->ia_size);
+                if (ret) {
+                        btrfs_truncate(inode);
+                        return ret;
+                }
+
+                i_size_write(inode, attr->ia_size);
+                btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
+
+                trans = btrfs_start_transaction(root, 1);
+                btrfs_set_trans_block_group(trans, inode);
+
+                ret = btrfs_update_inode(trans, root, inode);
+                BUG_ON(ret);
+                if (inode->i_nlink > 0) {
+                        ret = btrfs_orphan_del(trans, inode);
+                        BUG_ON(ret);
+                }
+                nr = trans->blocks_used;
+                btrfs_end_transaction(trans, root);
+                btrfs_btree_balance_dirty(root, nr);
+                return 0;
+        }
+
+        /*
+         * We're truncating a file that used to have good data down to
+         * zero. Make sure it gets into the ordered flush list so that
+         * any new writes get down to disk quickly.
+         */
+        if (attr->ia_size == 0)
+                BTRFS_I(inode)->ordered_data_close = 1;
+
+        /* we don't support swapfiles, so vmtruncate shouldn't fail */
+        ret = vmtruncate(inode, attr->ia_size);
+        BUG_ON(ret);
+
+        return 0;
+}
+
 static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
 {
         struct inode *inode = dentry->d_inode;
@@ -3179,23 +3331,14 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
                 return err;
 
         if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
-                if (attr->ia_size > inode->i_size) {
-                        err = btrfs_cont_expand(inode, attr->ia_size);
-                        if (err)
-                                return err;
-                } else if (inode->i_size > 0 &&
-                           attr->ia_size == 0) {
-
-                        /* we're truncating a file that used to have good
-                         * data down to zero.  Make sure it gets into
-                         * the ordered flush list so that any new writes
-                         * get down to disk quickly.
-                         */
-                        BTRFS_I(inode)->ordered_data_close = 1;
-                }
+                err = btrfs_setattr_size(inode, attr);
+                if (err)
+                        return err;
         }
+        attr->ia_valid &= ~ATTR_SIZE;
 
-        err = inode_setattr(inode, attr);
+        if (attr->ia_valid)
+                err = inode_setattr(inode, attr);
 
         if (!err && ((attr->ia_valid & ATTR_MODE)))
                 err = btrfs_acl_chmod(inode);
@@ -3216,36 +3359,43 @@ void btrfs_delete_inode(struct inode *inode)
         }
         btrfs_wait_ordered_range(inode, 0, (u64)-1);
 
+        if (root->fs_info->log_root_recovering) {
+                BUG_ON(!list_empty(&BTRFS_I(inode)->i_orphan));
+                goto no_delete;
+        }
+
         if (inode->i_nlink > 0) {
                 BUG_ON(btrfs_root_refs(&root->root_item) != 0);
                 goto no_delete;
         }
 
         btrfs_i_size_write(inode, 0);
-        trans = btrfs_join_transaction(root, 1);
 
-        btrfs_set_trans_block_group(trans, inode);
-        ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, 0);
-        if (ret) {
-                btrfs_orphan_del(NULL, inode);
-                goto no_delete_lock;
-        }
+        while (1) {
+                trans = btrfs_start_transaction(root, 1);
+                btrfs_set_trans_block_group(trans, inode);
+                ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
 
-        btrfs_orphan_del(trans, inode);
+                if (ret != -EAGAIN)
+                        break;
 
-        nr = trans->blocks_used;
-        clear_inode(inode);
+                nr = trans->blocks_used;
+                btrfs_end_transaction(trans, root);
+                trans = NULL;
+                btrfs_btree_balance_dirty(root, nr);
+        }
 
-        btrfs_end_transaction(trans, root);
-        btrfs_btree_balance_dirty(root, nr);
-        return;
+        if (ret == 0) {
+                ret = btrfs_orphan_del(trans, inode);
+                BUG_ON(ret);
+        }
 
-no_delete_lock:
         nr = trans->blocks_used;
         btrfs_end_transaction(trans, root);
         btrfs_btree_balance_dirty(root, nr);
 no_delete:
         clear_inode(inode);
+        return;
 }
 
 /*
@@ -3480,6 +3630,7 @@ static noinline void init_btrfs_i(struct inode *inode)
         bi->generation = 0;
         bi->sequence = 0;
         bi->last_trans = 0;
+        bi->last_sub_trans = 0;
         bi->logged_trans = 0;
         bi->delalloc_bytes = 0;
         bi->reserved_bytes = 0;
@@ -3497,7 +3648,6 @@ static noinline void init_btrfs_i(struct inode *inode)
         INIT_LIST_HEAD(&BTRFS_I(inode)->ordered_operations);
         RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
         btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
-        mutex_init(&BTRFS_I(inode)->extent_mutex);
         mutex_init(&BTRFS_I(inode)->log_mutex);
 }
 
@@ -3623,6 +3773,13 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
         }
         srcu_read_unlock(&root->fs_info->subvol_srcu, index);
 
+        if (root != sub_root) {
+                down_read(&root->fs_info->cleanup_work_sem);
+                if (!(inode->i_sb->s_flags & MS_RDONLY))
+                        btrfs_orphan_cleanup(sub_root);
+                up_read(&root->fs_info->cleanup_work_sem);
+        }
+
         return inode;
 }
 
@@ -3797,7 +3954,11 @@ skip:
 
         /* Reached end of directory/root. Bump pos past the last item. */
         if (key_type == BTRFS_DIR_INDEX_KEY)
-                filp->f_pos = INT_LIMIT(off_t);
+                /*
+                 * 32-bit glibc will use getdents64, but then strtol -
+                 * so the last number we can serve is this.
+                 */
+                filp->f_pos = 0x7fffffff;
         else
                 filp->f_pos++;
 nopos:
@@ -4147,7 +4308,7 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
         if (IS_ERR(inode))
                 goto out_unlock;
 
-        err = btrfs_init_inode_security(inode, dir);
+        err = btrfs_init_inode_security(trans, inode, dir);
         if (err) {
                 drop_inode = 1;
                 goto out_unlock;
@@ -4218,7 +4379,7 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
         if (IS_ERR(inode))
                 goto out_unlock;
 
-        err = btrfs_init_inode_security(inode, dir);
+        err = btrfs_init_inode_security(trans, inode, dir);
         if (err) {
                 drop_inode = 1;
                 goto out_unlock;
@@ -4264,6 +4425,10 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
         if (inode->i_nlink == 0)
                 return -ENOENT;
 
+        /* do not allow sys_link's with other subvols of the same device */
+        if (root->objectid != BTRFS_I(inode)->root->objectid)
+                return -EPERM;
+
         /*
          * 1 item for inode ref
          * 2 items for dir items
@@ -4351,7 +4516,7 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 
         drop_on_err = 1;
 
-        err = btrfs_init_inode_security(inode, dir);
+        err = btrfs_init_inode_security(trans, inode, dir);
         if (err)
                 goto out_fail;
 
@@ -4980,7 +5145,9 @@ again:
         set_page_dirty(page);
         SetPageUptodate(page);
 
-        BTRFS_I(inode)->last_trans = root->fs_info->generation + 1;
+        BTRFS_I(inode)->last_trans = root->fs_info->generation;
+        BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
+
         unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 
 out_unlock:
@@ -5000,15 +5167,20 @@ static void btrfs_truncate(struct inode *inode)
         unsigned long nr;
         u64 mask = root->sectorsize - 1;
 
-        if (!S_ISREG(inode->i_mode))
+        if (!S_ISREG(inode->i_mode)) {
+                WARN_ON(1);
                 return;
-        if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
+        }
+
+        ret = btrfs_truncate_page(inode->i_mapping, inode->i_size);
+        if (ret)
                 return;
 
-        btrfs_truncate_page(inode->i_mapping, inode->i_size);
         btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
+        btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
 
         trans = btrfs_start_transaction(root, 1);
+        btrfs_set_trans_block_group(trans, inode);
 
         /*
          * setattr is responsible for setting the ordered_data_close flag,
@@ -5030,21 +5202,32 @@ static void btrfs_truncate(struct inode *inode)
         if (inode->i_size == 0 && BTRFS_I(inode)->ordered_data_close)
                 btrfs_add_ordered_operation(trans, root, inode);
 
-        btrfs_set_trans_block_group(trans, inode);
-        btrfs_i_size_write(inode, inode->i_size);
+        while (1) {
+                ret = btrfs_truncate_inode_items(trans, root, inode,
+                                                 inode->i_size,
+                                                 BTRFS_EXTENT_DATA_KEY);
+                if (ret != -EAGAIN)
+                        break;
 
-        ret = btrfs_orphan_add(trans, inode);
-        if (ret)
-                goto out;
-        /* FIXME, add redo link to tree so we don't leak on crash */
-        ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size,
-                                      BTRFS_EXTENT_DATA_KEY);
-        btrfs_update_inode(trans, root, inode);
+                ret = btrfs_update_inode(trans, root, inode);
+                BUG_ON(ret);
+
+                nr = trans->blocks_used;
+                btrfs_end_transaction(trans, root);
+                btrfs_btree_balance_dirty(root, nr);
 
-        ret = btrfs_orphan_del(trans, inode);
+                trans = btrfs_start_transaction(root, 1);
+                btrfs_set_trans_block_group(trans, inode);
+        }
+
+        if (ret == 0 && inode->i_nlink > 0) {
+                ret = btrfs_orphan_del(trans, inode);
+                BUG_ON(ret);
+        }
+
+        ret = btrfs_update_inode(trans, root, inode);
         BUG_ON(ret);
 
-out:
         nr = trans->blocks_used;
         ret = btrfs_end_transaction_throttle(trans, root);
         BUG_ON(ret);
@@ -5100,9 +5283,11 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
         if (!ei)
                 return NULL;
         ei->last_trans = 0;
+        ei->last_sub_trans = 0;
         ei->logged_trans = 0;
         ei->outstanding_extents = 0;
         ei->reserved_extents = 0;
+        ei->root = NULL;
         spin_lock_init(&ei->accounting_lock);
         btrfs_ordered_inode_tree_init(&ei->ordered_tree);
         INIT_LIST_HEAD(&ei->i_orphan);
@@ -5119,6 +5304,14 @@ void btrfs_destroy_inode(struct inode *inode)
         WARN_ON(inode->i_data.nrpages);
 
         /*
+         * This can happen where we create an inode, but somebody else also
+         * created the same inode and we need to destroy the one we already
+         * created.
+         */
+        if (!root)
+                goto free;
+
+        /*
          * Make sure we're properly removed from the ordered operation
          * lists.
          */
@@ -5131,9 +5324,9 @@ void btrfs_destroy_inode(struct inode *inode)
 
         spin_lock(&root->list_lock);
         if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
-                printk(KERN_ERR "BTRFS: inode %lu: inode still on the orphan"
-                       " list\n", inode->i_ino);
-                dump_stack();
+                printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n",
+                       inode->i_ino);
+                list_del_init(&BTRFS_I(inode)->i_orphan);
         }
         spin_unlock(&root->list_lock);
 
@@ -5153,6 +5346,7 @@ void btrfs_destroy_inode(struct inode *inode)
         }
         inode_tree_del(inode);
         btrfs_drop_extent_cache(inode, 0, (u64)-1, 0);
+free:
         kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
 }
 
@@ -5258,11 +5452,14 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
                 return -ENOTEMPTY;
 
         /*
-         * 2 items for dir items
-         * 1 item for orphan entry
-         * 1 item for ref
+         * We want to reserve the absolute worst case amount of items.  So if
+         * both inodes are subvols and we need to unlink them then that would
+         * require 4 item modifications, but if they are both normal inodes it
+         * would require 5 item modifications, so we'll assume their normal
+         * inodes.  So 5 * 2 is 10, plus 1 for the new link, so 11 total items
+         * should cover the worst case number of items we'll modify.
          */
-        ret = btrfs_reserve_metadata_space(root, 4);
+        ret = btrfs_reserve_metadata_space(root, 11);
         if (ret)
                 return ret;
 
@@ -5378,7 +5575,7 @@ out_fail:
         if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
                 up_read(&root->fs_info->subvol_sem);
 
-        btrfs_unreserve_metadata_space(root, 4);
+        btrfs_unreserve_metadata_space(root, 11);
         return ret;
 }
 
@@ -5386,7 +5583,7 @@ out_fail:
  * some fairly slow code that needs optimization. This walks the list
  * of all the inodes with pending delalloc and forces them to disk.
  */
-int btrfs_start_delalloc_inodes(struct btrfs_root *root)
+int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput)
 {
         struct list_head *head = &root->fs_info->delalloc_inodes;
         struct btrfs_inode *binode;
@@ -5405,7 +5602,10 @@ int btrfs_start_delalloc_inodes(struct btrfs_root *root)
                 spin_unlock(&root->fs_info->delalloc_lock);
                 if (inode) {
                         filemap_flush(inode->i_mapping);
-                        iput(inode);
+                        if (delay_iput)
+                                btrfs_add_delayed_iput(inode);
+                        else
+                                iput(inode);
                 }
                 cond_resched();
                 spin_lock(&root->fs_info->delalloc_lock);
@@ -5479,7 +5679,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
         if (IS_ERR(inode))
                 goto out_unlock;
 
-        err = btrfs_init_inode_security(inode, dir);
+        err = btrfs_init_inode_security(trans, inode, dir);
         if (err) {
                 drop_inode = 1;
                 goto out_unlock;
@@ -5551,57 +5751,77 @@ out_fail:
         return err;
 }
 
-static int prealloc_file_range(struct btrfs_trans_handle *trans,
-                               struct inode *inode, u64 start, u64 end,
-                               u64 locked_end, u64 alloc_hint, int mode)
+static int prealloc_file_range(struct inode *inode, u64 start, u64 end,
+                        u64 alloc_hint, int mode, loff_t actual_len)
 {
+        struct btrfs_trans_handle *trans;
         struct btrfs_root *root = BTRFS_I(inode)->root;
         struct btrfs_key ins;
         u64 alloc_size;
         u64 cur_offset = start;
         u64 num_bytes = end - start;
         int ret = 0;
+        u64 i_size;
 
         while (num_bytes > 0) {
                 alloc_size = min(num_bytes, root->fs_info->max_extent);
 
-                ret = btrfs_reserve_metadata_space(root, 1);
-                if (ret)
-                        goto out;
+                trans = btrfs_start_transaction(root, 1);
 
                 ret = btrfs_reserve_extent(trans, root, alloc_size,
                                            root->sectorsize, 0, alloc_hint,
                                            (u64)-1, &ins, 1);
                 if (ret) {
                         WARN_ON(1);
-                        goto out;
+                        goto stop_trans;
+                }
+
+                ret = btrfs_reserve_metadata_space(root, 3);
+                if (ret) {
+                        btrfs_free_reserved_extent(root, ins.objectid,
+                                                   ins.offset);
+                        goto stop_trans;
                 }
+
                 ret = insert_reserved_file_extent(trans, inode,
                                                   cur_offset, ins.objectid,
                                                   ins.offset, ins.offset,
-                                                  ins.offset, locked_end,
-                                                  0, 0, 0,
+                                                  ins.offset, 0, 0, 0,
                                                   BTRFS_FILE_EXTENT_PREALLOC);
                 BUG_ON(ret);
                 btrfs_drop_extent_cache(inode, cur_offset,
                                         cur_offset + ins.offset -1, 0);
+
                 num_bytes -= ins.offset;
                 cur_offset += ins.offset;
                 alloc_hint = ins.objectid + ins.offset;
-                btrfs_unreserve_metadata_space(root, 1);
-        }
-out:
-        if (cur_offset > start) {
+
                 inode->i_ctime = CURRENT_TIME;
                 BTRFS_I(inode)->flags |= BTRFS_INODE_PREALLOC;
                 if (!(mode & FALLOC_FL_KEEP_SIZE) &&
-                    cur_offset > i_size_read(inode))
-                        btrfs_i_size_write(inode, cur_offset);
+                        (actual_len > inode->i_size) &&
+                        (cur_offset > inode->i_size)) {
+
+                        if (cur_offset > actual_len)
+                                i_size  = actual_len;
+                        else
+                                i_size = cur_offset;
+                        i_size_write(inode, i_size);
+                        btrfs_ordered_update_i_size(inode, i_size, NULL);
+                }
+
                 ret = btrfs_update_inode(trans, root, inode);
                 BUG_ON(ret);
+
+                btrfs_end_transaction(trans, root);
+                btrfs_unreserve_metadata_space(root, 3);
         }
+        return ret;
 
+stop_trans:
+        btrfs_end_transaction(trans, root);
         return ret;
+
 }
 
 static long btrfs_fallocate(struct inode *inode, int mode,
@@ -5615,8 +5835,6 @@ static long btrfs_fallocate(struct inode *inode, int mode,
         u64 locked_end;
         u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
         struct extent_map *em;
-        struct btrfs_trans_handle *trans;
-        struct btrfs_root *root;
         int ret;
 
         alloc_start = offset & ~mask;
@@ -5635,9 +5853,7 @@ static long btrfs_fallocate(struct inode *inode, int mode,
                         goto out;
         }
 
-        root = BTRFS_I(inode)->root;
-
-        ret = btrfs_check_data_free_space(root, inode,
+        ret = btrfs_check_data_free_space(BTRFS_I(inode)->root, inode,
                                           alloc_end - alloc_start);
         if (ret)
                 goto out;
@@ -5646,12 +5862,6 @@ static long btrfs_fallocate(struct inode *inode, int mode,
         while (1) {
                 struct btrfs_ordered_extent *ordered;
 
-                trans = btrfs_start_transaction(BTRFS_I(inode)->root, 1);
-                if (!trans) {
-                        ret = -EIO;
-                        goto out_free;
-                }
-
                 /* the extent lock is ordered inside the running
                  * transaction
                  */
@@ -5665,8 +5875,6 @@ static long btrfs_fallocate(struct inode *inode, int mode,
                         btrfs_put_ordered_extent(ordered);
                         unlock_extent(&BTRFS_I(inode)->io_tree,
                                       alloc_start, locked_end, GFP_NOFS);
-                        btrfs_end_transaction(trans, BTRFS_I(inode)->root);
-
                         /*
                          * we can't wait on the range with the transaction
                          * running or with the extent lock held
@@ -5687,10 +5895,12 @@ static long btrfs_fallocate(struct inode *inode, int mode,
                 BUG_ON(IS_ERR(em) || !em);
                 last_byte = min(extent_map_end(em), alloc_end);
                 last_byte = (last_byte + mask) & ~mask;
-                if (em->block_start == EXTENT_MAP_HOLE) {
-                        ret = prealloc_file_range(trans, inode, cur_offset,
-                                        last_byte, locked_end + 1,
-                                        alloc_hint, mode);
+                if (em->block_start == EXTENT_MAP_HOLE ||
+                    (cur_offset >= inode->i_size &&
+                     !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
+                        ret = prealloc_file_range(inode,
+                                                  cur_offset, last_byte,
+                                                alloc_hint, mode, offset+len);
                         if (ret < 0) {
                                 free_extent_map(em);
                                 break;
@@ -5709,9 +5919,8 @@ static long btrfs_fallocate(struct inode *inode, int mode,
         unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
                       GFP_NOFS);
 
-        btrfs_end_transaction(trans, BTRFS_I(inode)->root);
-out_free:
-        btrfs_free_reserved_data_space(root, inode, alloc_end - alloc_start);
+        btrfs_free_reserved_data_space(BTRFS_I(inode)->root, inode,
+                                       alloc_end - alloc_start);
 out:
         mutex_unlock(&inode->i_mutex);
         return ret;
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index cdbb054102b9..645a17927a8f 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -237,7 +237,6 @@ static noinline int create_subvol(struct btrfs_root *root,
         u64 objectid;
         u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
         u64 index = 0;
-        unsigned long nr = 1;
 
         /*
          * 1 - inode item
@@ -290,7 +289,7 @@ static noinline int create_subvol(struct btrfs_root *root,
         btrfs_set_root_generation(&root_item, trans->transid);
         btrfs_set_root_level(&root_item, 0);
         btrfs_set_root_refs(&root_item, 1);
-        btrfs_set_root_used(&root_item, 0);
+        btrfs_set_root_used(&root_item, leaf->len);
         btrfs_set_root_last_snapshot(&root_item, 0);
 
         memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
@@ -342,24 +341,21 @@ static noinline int create_subvol(struct btrfs_root *root,
 
         d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
 fail:
-        nr = trans->blocks_used;
         err = btrfs_commit_transaction(trans, root);
         if (err && !ret)
                 ret = err;
 
         btrfs_unreserve_metadata_space(root, 6);
-        btrfs_btree_balance_dirty(root, nr);
         return ret;
 }
 
 static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
                            char *name, int namelen)
 {
+        struct inode *inode;
         struct btrfs_pending_snapshot *pending_snapshot;
         struct btrfs_trans_handle *trans;
-        int ret = 0;
-        int err;
-        unsigned long nr = 0;
+        int ret;
 
         if (!root->ref_cows)
                 return -EINVAL;
@@ -372,20 +368,20 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
          */
         ret = btrfs_reserve_metadata_space(root, 6);
         if (ret)
-                goto fail_unlock;
+                goto fail;
 
         pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
         if (!pending_snapshot) {
                 ret = -ENOMEM;
                 btrfs_unreserve_metadata_space(root, 6);
-                goto fail_unlock;
+                goto fail;
         }
         pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
         if (!pending_snapshot->name) {
                 ret = -ENOMEM;
                 kfree(pending_snapshot);
                 btrfs_unreserve_metadata_space(root, 6);
-                goto fail_unlock;
+                goto fail;
         }
         memcpy(pending_snapshot->name, name, namelen);
         pending_snapshot->name[namelen] = '\0';
@@ -395,10 +391,19 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
         pending_snapshot->root = root;
         list_add(&pending_snapshot->list,
                  &trans->transaction->pending_snapshots);
-        err = btrfs_commit_transaction(trans, root);
+        ret = btrfs_commit_transaction(trans, root);
+        BUG_ON(ret);
+        btrfs_unreserve_metadata_space(root, 6);
 
-fail_unlock:
-        btrfs_btree_balance_dirty(root, nr);
+        inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
+        if (IS_ERR(inode)) {
+                ret = PTR_ERR(inode);
+                goto fail;
+        }
+        BUG_ON(!inode);
+        d_instantiate(dentry, inode);
+        ret = 0;
+fail:
         return ret;
 }
 
@@ -1027,8 +1032,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
         BUG_ON(!trans);
 
         /* punch hole in destination first */
-        btrfs_drop_extents(trans, root, inode, off, off + len,
-                           off + len, 0, &hint_byte, 1);
+        btrfs_drop_extents(trans, inode, off, off + len, &hint_byte, 1);
 
         /* clone data */
         key.objectid = src->i_ino;
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 5799bc46a309..5c2a9e78a949 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -291,16 +291,16 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 
 /*
  * remove an ordered extent from the tree.  No references are dropped
- * but, anyone waiting on this extent is woken up.
+ * and you must wake_up entry->wait.  You must hold the tree mutex
+ * while you call this function.
  */
-int btrfs_remove_ordered_extent(struct inode *inode,
+static int __btrfs_remove_ordered_extent(struct inode *inode,
                                 struct btrfs_ordered_extent *entry)
 {
         struct btrfs_ordered_inode_tree *tree;
         struct rb_node *node;
 
         tree = &BTRFS_I(inode)->ordered_tree;
-        mutex_lock(&tree->mutex);
         node = &entry->rb_node;
         rb_erase(node, &tree->tree);
         tree->last = NULL;
@@ -326,16 +326,34 @@ int btrfs_remove_ordered_extent(struct inode *inode,
         }
         spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
 
+        return 0;
+}
+
+/*
+ * remove an ordered extent from the tree.  No references are dropped
+ * but any waiters are woken.
+ */
+int btrfs_remove_ordered_extent(struct inode *inode,
+                                struct btrfs_ordered_extent *entry)
+{
+        struct btrfs_ordered_inode_tree *tree;
+        int ret;
+
+        tree = &BTRFS_I(inode)->ordered_tree;
+        mutex_lock(&tree->mutex);
+        ret = __btrfs_remove_ordered_extent(inode, entry);
         mutex_unlock(&tree->mutex);
         wake_up(&entry->wait);
-        return 0;
+
+        return ret;
 }
 
 /*
  * wait for all the ordered extents in a root.  This is done when balancing
  * space between drives.
  */
-int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only)
+int btrfs_wait_ordered_extents(struct btrfs_root *root,
+                               int nocow_only, int delay_iput)
 {
         struct list_head splice;
         struct list_head *cur;
@@ -372,7 +390,10 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only)
                 if (inode) {
                         btrfs_start_ordered_extent(inode, ordered, 1);
                         btrfs_put_ordered_extent(ordered);
-                        iput(inode);
+                        if (delay_iput)
+                                btrfs_add_delayed_iput(inode);
+                        else
+                                iput(inode);
                 } else {
                         btrfs_put_ordered_extent(ordered);
                 }
@@ -430,7 +451,7 @@ again:
                                 btrfs_wait_ordered_range(inode, 0, (u64)-1);
                         else
                                 filemap_flush(inode->i_mapping);
-                        iput(inode);
+                        btrfs_add_delayed_iput(inode);
                 }
 
                 cond_resched();
@@ -589,7 +610,7 @@ out:
  * After an extent is done, call this to conditionally update the on disk
  * i_size.  i_size is updated to cover any fully written part of the file.
  */
-int btrfs_ordered_update_i_size(struct inode *inode,
+int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
                                 struct btrfs_ordered_extent *ordered)
 {
         struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
@@ -597,18 +618,32 @@ int btrfs_ordered_update_i_size(struct inode *inode,
         u64 disk_i_size;
         u64 new_i_size;
         u64 i_size_test;
+        u64 i_size = i_size_read(inode);
         struct rb_node *node;
+        struct rb_node *prev = NULL;
         struct btrfs_ordered_extent *test;
+        int ret = 1;
+
+        if (ordered)
+                offset = entry_end(ordered);
+        else
+                offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize);
 
         mutex_lock(&tree->mutex);
         disk_i_size = BTRFS_I(inode)->disk_i_size;
 
+        /* truncate file */
+        if (disk_i_size > i_size) {
+                BTRFS_I(inode)->disk_i_size = i_size;
+                ret = 0;
+                goto out;
+        }
+
         /*
          * if the disk i_size is already at the inode->i_size, or
          * this ordered extent is inside the disk i_size, we're done
          */
-        if (disk_i_size >= inode->i_size ||
-            ordered->file_offset + ordered->len <= disk_i_size) {
+        if (disk_i_size == i_size || offset <= disk_i_size) {
                 goto out;
         }
 
@@ -616,8 +651,7 @@ int btrfs_ordered_update_i_size(struct inode *inode,
          * we can't update the disk_isize if there are delalloc bytes
          * between disk_i_size and  this ordered extent
          */
-        if (test_range_bit(io_tree, disk_i_size,
-                           ordered->file_offset + ordered->len - 1,
+        if (test_range_bit(io_tree, disk_i_size, offset - 1,
                            EXTENT_DELALLOC, 0, NULL)) {
                 goto out;
         }
@@ -626,20 +660,32 @@ int btrfs_ordered_update_i_size(struct inode *inode,
          * if we find an ordered extent then we can't update disk i_size
          * yet
          */
-        node = &ordered->rb_node;
-        while (1) {
-                node = rb_prev(node);
-                if (!node)
-                        break;
+        if (ordered) {
+                node = rb_prev(&ordered->rb_node);
+        } else {
+                prev = tree_search(tree, offset);
+                /*
+                 * we insert file extents without involving ordered struct,
+                 * so there should be no ordered struct cover this offset
+                 */
+                if (prev) {
+                        test = rb_entry(prev, struct btrfs_ordered_extent,
+                                        rb_node);
+                        BUG_ON(offset_in_entry(test, offset));
+                }
+                node = prev;
+        }
+        while (node) {
                 test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
                 if (test->file_offset + test->len <= disk_i_size)
                         break;
-                if (test->file_offset >= inode->i_size)
+                if (test->file_offset >= i_size)
                         break;
                 if (test->file_offset >= disk_i_size)
                         goto out;
+                node = rb_prev(node);
         }
-        new_i_size = min_t(u64, entry_end(ordered), i_size_read(inode));
+        new_i_size = min_t(u64, offset, i_size);
 
         /*
          * at this point, we know we can safely update i_size to at least
@@ -647,7 +693,14 @@ int btrfs_ordered_update_i_size(struct inode *inode,
          * walk forward and see if ios from higher up in the file have
          * finished.
          */
-        node = rb_next(&ordered->rb_node);
+        if (ordered) {
+                node = rb_next(&ordered->rb_node);
+        } else {
+                if (prev)
+                        node = rb_next(prev);
+                else
+                        node = rb_first(&tree->tree);
+        }
         i_size_test = 0;
         if (node) {
                 /*
@@ -655,10 +708,10 @@ int btrfs_ordered_update_i_size(struct inode *inode,
                  * between our ordered extent and the next one.
                  */
                 test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-                if (test->file_offset > entry_end(ordered))
+                if (test->file_offset > offset)
                         i_size_test = test->file_offset;
         } else {
-                i_size_test = i_size_read(inode);
+                i_size_test = i_size;
         }
 
         /*
@@ -667,15 +720,25 @@ int btrfs_ordered_update_i_size(struct inode *inode,
          * are no delalloc bytes in this area, it is safe to update
          * disk_i_size to the end of the region.
          */
-        if (i_size_test > entry_end(ordered) &&
-            !test_range_bit(io_tree, entry_end(ordered), i_size_test - 1,
-                           EXTENT_DELALLOC, 0, NULL)) {
-                new_i_size = min_t(u64, i_size_test, i_size_read(inode));
+        if (i_size_test > offset &&
+            !test_range_bit(io_tree, offset, i_size_test - 1,
+                            EXTENT_DELALLOC, 0, NULL)) {
+                new_i_size = min_t(u64, i_size_test, i_size);
         }
         BTRFS_I(inode)->disk_i_size = new_i_size;
+        ret = 0;
 out:
+        /*
+         * we need to remove the ordered extent with the tree lock held
+         * so that other people calling this function don't find our fully
+         * processed ordered entry and skip updating the i_size
+         */
+        if (ordered)
+                __btrfs_remove_ordered_extent(inode, ordered);
         mutex_unlock(&tree->mutex);
-        return 0;
+        if (ordered)
+                wake_up(&ordered->wait);
+        return ret;
 }
 
 /*
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index f82e87488ca8..1fe1282ef47c 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -150,12 +150,13 @@ void btrfs_start_ordered_extent(struct inode *inode,
 int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
 struct btrfs_ordered_extent *
 btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
-int btrfs_ordered_update_i_size(struct inode *inode,
+int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
                                 struct btrfs_ordered_extent *ordered);
 int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum);
-int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only);
 int btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
 int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
                                 struct btrfs_root *root,
                                 struct inode *inode);
+int btrfs_wait_ordered_extents(struct btrfs_root *root,
+                               int nocow_only, int delay_iput);
 #endif
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index cfcc93c93a7b..ab7ab5318745 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -1561,6 +1561,20 @@ static int invalidate_extent_cache(struct btrfs_root *root,
         return 0;
 }
 
+static void put_inodes(struct list_head *list)
+{
+        struct inodevec *ivec;
+        while (!list_empty(list)) {
+                ivec = list_entry(list->next, struct inodevec, list);
+                list_del(&ivec->list);
+                while (ivec->nr > 0) {
+                        ivec->nr--;
+                        iput(ivec->inode[ivec->nr]);
+                }
+                kfree(ivec);
+        }
+}
+
 static int find_next_key(struct btrfs_path *path, int level,
                          struct btrfs_key *key)
 
@@ -1723,6 +1737,11 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 
                 btrfs_btree_balance_dirty(root, nr);
 
+                /*
+                 * put inodes outside transaction, otherwise we may deadlock.
+                 */
+                put_inodes(&inode_list);
+
                 if (replaced && rc->stage == UPDATE_DATA_PTRS)
                         invalidate_extent_cache(root, &key, &next_key);
         }
@@ -1752,19 +1771,7 @@ out:
 
         btrfs_btree_balance_dirty(root, nr);
 
-        /*
-         * put inodes while we aren't holding the tree locks
-         */
-        while (!list_empty(&inode_list)) {
-                struct inodevec *ivec;
-                ivec = list_entry(inode_list.next, struct inodevec, list);
-                list_del(&ivec->list);
-                while (ivec->nr > 0) {
-                        ivec->nr--;
-                        iput(ivec->inode[ivec->nr]);
-                }
-                kfree(ivec);
-        }
+        put_inodes(&inode_list);
 
         if (replaced && rc->stage == UPDATE_DATA_PTRS)
                 invalidate_extent_cache(root, &key, &next_key);
@@ -3274,8 +3281,10 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
                 return -ENOMEM;
 
         path = btrfs_alloc_path();
-        if (!path)
+        if (!path) {
+                kfree(cluster);
                 return -ENOMEM;
+        }
 
         rc->extents_found = 0;
         rc->extents_skipped = 0;
@@ -3534,8 +3543,8 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
                (unsigned long long)rc->block_group->key.objectid,
                (unsigned long long)rc->block_group->flags);
 
-        btrfs_start_delalloc_inodes(fs_info->tree_root);
-        btrfs_wait_ordered_extents(fs_info->tree_root, 0);
+        btrfs_start_delalloc_inodes(fs_info->tree_root, 0);
+        btrfs_wait_ordered_extents(fs_info->tree_root, 0, 0);
 
         while (1) {
                 rc->extents_found = 0;
@@ -3755,6 +3764,8 @@ out:
                                        BTRFS_DATA_RELOC_TREE_OBJECTID);
                 if (IS_ERR(fs_root))
                         err = PTR_ERR(fs_root);
+                else
+                        btrfs_orphan_cleanup(fs_root);
         }
         return err;
 }
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index 9351428f30e2..67fa2d29d663 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -159,7 +159,6 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
         write_extent_buffer(l, item, ptr, sizeof(*item));
         btrfs_mark_buffer_dirty(path->nodes[0]);
 out:
-        btrfs_release_path(root, path);
         btrfs_free_path(path);
         return ret;
 }
@@ -332,7 +331,6 @@ int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
         BUG_ON(refs != 0);
         ret = btrfs_del_item(trans, root, path);
 out:
-        btrfs_release_path(root, path);
         btrfs_free_path(path);
         return ret;
 }
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 9de9b2236419..8a1ea6e64575 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -66,7 +66,9 @@ enum {
         Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
         Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
         Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl,
-        Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_err,
+        Opt_compress, Opt_compress_force, Opt_notreelog, Opt_ratio,
+        Opt_flushoncommit,
+        Opt_discard, Opt_err,
 };
 
 static match_table_t tokens = {
@@ -81,6 +83,7 @@ static match_table_t tokens = {
         {Opt_alloc_start, "alloc_start=%s"},
         {Opt_thread_pool, "thread_pool=%d"},
         {Opt_compress, "compress"},
+        {Opt_compress_force, "compress-force"},
         {Opt_ssd, "ssd"},
         {Opt_ssd_spread, "ssd_spread"},
         {Opt_nossd, "nossd"},
@@ -88,6 +91,7 @@ static match_table_t tokens = {
         {Opt_notreelog, "notreelog"},
         {Opt_flushoncommit, "flushoncommit"},
         {Opt_ratio, "metadata_ratio=%d"},
+        {Opt_discard, "discard"},
         {Opt_err, NULL},
 };
 
@@ -126,6 +130,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
         substring_t args[MAX_OPT_ARGS];
         char *p, *num;
         int intarg;
+        int ret = 0;
 
         if (!options)
                 return 0;
@@ -170,6 +175,11 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
                         printk(KERN_INFO "btrfs: use compression\n");
                         btrfs_set_opt(info->mount_opt, COMPRESS);
                         break;
+                case Opt_compress_force:
+                        printk(KERN_INFO "btrfs: forcing compression\n");
+                        btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
+                        btrfs_set_opt(info->mount_opt, COMPRESS);
+                        break;
                 case Opt_ssd:
                         printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
                         btrfs_set_opt(info->mount_opt, SSD);
@@ -257,12 +267,21 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
                                        info->metadata_ratio);
                         }
                         break;
+                case Opt_discard:
+                        btrfs_set_opt(info->mount_opt, DISCARD);
+                        break;
+                case Opt_err:
+                        printk(KERN_INFO "btrfs: unrecognized mount option "
+                               "'%s'\n", p);
+                        ret = -EINVAL;
+                        goto out;
                 default:
                         break;
                 }
         }
+out:
         kfree(options);
-        return 0;
+        return ret;
 }
 
 /*
@@ -344,7 +363,7 @@ static int btrfs_fill_super(struct super_block *sb,
         sb->s_export_op = &btrfs_export_ops;
         sb->s_xattr = btrfs_xattr_handlers;
         sb->s_time_gran = 1;
-#ifdef CONFIG_BTRFS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
         sb->s_flags |= MS_POSIXACL;
 #endif
 
@@ -400,8 +419,8 @@ int btrfs_sync_fs(struct super_block *sb, int wait)
                 return 0;
         }
 
-        btrfs_start_delalloc_inodes(root);
-        btrfs_wait_ordered_extents(root, 0);
+        btrfs_start_delalloc_inodes(root, 0);
+        btrfs_wait_ordered_extents(root, 0, 0);
 
         trans = btrfs_start_transaction(root, 1);
         ret = btrfs_commit_transaction(trans, root);
@@ -445,6 +464,8 @@ static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
                 seq_puts(seq, ",notreelog");
         if (btrfs_test_opt(root, FLUSHONCOMMIT))
                 seq_puts(seq, ",flushoncommit");
+        if (btrfs_test_opt(root, DISCARD))
+                seq_puts(seq, ",discard");
         if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
                 seq_puts(seq, ",noacl");
         return 0;
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 0b8f36d4400a..b2acc79f1b34 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -163,8 +163,14 @@ static void wait_current_trans(struct btrfs_root *root)
         }
 }
 
+enum btrfs_trans_type {
+        TRANS_START,
+        TRANS_JOIN,
+        TRANS_USERSPACE,
+};
+
 static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
-                                             int num_blocks, int wait)
+                                             int num_blocks, int type)
 {
         struct btrfs_trans_handle *h =
                 kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
@@ -172,7 +178,8 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 
         mutex_lock(&root->fs_info->trans_mutex);
         if (!root->fs_info->log_root_recovering &&
-            ((wait == 1 && !root->fs_info->open_ioctl_trans) || wait == 2))
+            ((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
+             type == TRANS_USERSPACE))
                 wait_current_trans(root);
         ret = join_transaction(root);
         BUG_ON(ret);
@@ -186,7 +193,7 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
         h->alloc_exclude_start = 0;
         h->delayed_ref_updates = 0;
 
-        if (!current->journal_info)
+        if (!current->journal_info && type != TRANS_USERSPACE)
                 current->journal_info = h;
 
         root->fs_info->running_transaction->use_count++;
@@ -198,18 +205,18 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
                                                    int num_blocks)
 {
-        return start_transaction(root, num_blocks, 1);
+        return start_transaction(root, num_blocks, TRANS_START);
 }
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
                                                    int num_blocks)
 {
-        return start_transaction(root, num_blocks, 0);
+        return start_transaction(root, num_blocks, TRANS_JOIN);
 }
 
 struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
                                                          int num_blocks)
 {
-        return start_transaction(r, num_blocks, 2);
+        return start_transaction(r, num_blocks, TRANS_USERSPACE);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -326,6 +333,9 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
         memset(trans, 0, sizeof(*trans));
         kmem_cache_free(btrfs_trans_handle_cachep, trans);
 
+        if (throttle)
+                btrfs_run_delayed_iputs(root);
+
         return 0;
 }
 
@@ -344,10 +354,10 @@ int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans,
 /*
  * when btree blocks are allocated, they have some corresponding bits set for
  * them in one of two extent_io trees.  This is used to make sure all of
- * those extents are on disk for transaction or log commit
+ * those extents are sent to disk but does not wait on them
  */
-int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
-                                        struct extent_io_tree *dirty_pages)
+int btrfs_write_marked_extents(struct btrfs_root *root,
+                               struct extent_io_tree *dirty_pages, int mark)
 {
         int ret;
         int err = 0;
@@ -360,7 +370,7 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
 
         while (1) {
                 ret = find_first_extent_bit(dirty_pages, start, &start, &end,
-                                            EXTENT_DIRTY);
+                                            mark);
                 if (ret)
                         break;
                 while (start <= end) {
@@ -394,13 +404,36 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
                         page_cache_release(page);
                 }
         }
+        if (err)
+                werr = err;
+        return werr;
+}
+
+/*
+ * when btree blocks are allocated, they have some corresponding bits set for
+ * them in one of two extent_io trees.  This is used to make sure all of
+ * those extents are on disk for transaction or log commit.  We wait
+ * on all the pages and clear them from the dirty pages state tree
+ */
+int btrfs_wait_marked_extents(struct btrfs_root *root,
+                              struct extent_io_tree *dirty_pages, int mark)
+{
+        int ret;
+        int err = 0;
+        int werr = 0;
+        struct page *page;
+        struct inode *btree_inode = root->fs_info->btree_inode;
+        u64 start = 0;
+        u64 end;
+        unsigned long index;
+
         while (1) {
-                ret = find_first_extent_bit(dirty_pages, 0, &start, &end,
-                                            EXTENT_DIRTY);
+                ret = find_first_extent_bit(dirty_pages, start, &start, &end,
+                                            mark);
                 if (ret)
                         break;
 
-                clear_extent_dirty(dirty_pages, start, end, GFP_NOFS);
+                clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
                 while (start <= end) {
                         index = start >> PAGE_CACHE_SHIFT;
                         start = (u64)(index + 1) << PAGE_CACHE_SHIFT;
@@ -424,6 +457,22 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
         return werr;
 }
 
+/*
+ * when btree blocks are allocated, they have some corresponding bits set for
+ * them in one of two extent_io trees.  This is used to make sure all of
+ * those extents are on disk for transaction or log commit
+ */
+int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
+                                struct extent_io_tree *dirty_pages, int mark)
+{
+        int ret;
+        int ret2;
+
+        ret = btrfs_write_marked_extents(root, dirty_pages, mark);
+        ret2 = btrfs_wait_marked_extents(root, dirty_pages, mark);
+        return ret || ret2;
+}
+
 int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
                                      struct btrfs_root *root)
 {
@@ -433,7 +482,8 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
                 return filemap_write_and_wait(btree_inode->i_mapping);
         }
         return btrfs_write_and_wait_marked_extents(root,
-                                           &trans->transaction->dirty_pages);
+                                           &trans->transaction->dirty_pages,
+                                           EXTENT_DIRTY);
 }
 
 /*
@@ -451,13 +501,16 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
 {
         int ret;
         u64 old_root_bytenr;
+        u64 old_root_used;
         struct btrfs_root *tree_root = root->fs_info->tree_root;
 
+        old_root_used = btrfs_root_used(&root->root_item);
         btrfs_write_dirty_block_groups(trans, root);
 
         while (1) {
                 old_root_bytenr = btrfs_root_bytenr(&root->root_item);
-                if (old_root_bytenr == root->node->start)
+                if (old_root_bytenr == root->node->start &&
+                    old_root_used == btrfs_root_used(&root->root_item))
                         break;
 
                 btrfs_set_root_node(&root->root_item, root->node);
@@ -466,6 +519,7 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
                                         &root->root_item);
                 BUG_ON(ret);
 
+                old_root_used = btrfs_root_used(&root->root_item);
                 ret = btrfs_write_dirty_block_groups(trans, root);
                 BUG_ON(ret);
         }
@@ -749,7 +803,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         memcpy(&pending->root_key, &key, sizeof(key));
 fail:
         kfree(new_root_item);
-        btrfs_unreserve_metadata_space(root, 6);
         return ret;
 }
 
@@ -761,7 +814,6 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info,
         u64 index = 0;
         struct btrfs_trans_handle *trans;
         struct inode *parent_inode;
-        struct inode *inode;
         struct btrfs_root *parent_root;
 
         parent_inode = pending->dentry->d_parent->d_inode;
@@ -793,8 +845,6 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info,
 
         BUG_ON(ret);
 
-        inode = btrfs_lookup_dentry(parent_inode, pending->dentry);
-        d_instantiate(pending->dentry, inode);
 fail:
         btrfs_end_transaction(trans, fs_info->fs_root);
         return ret;
@@ -948,11 +998,11 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                 mutex_unlock(&root->fs_info->trans_mutex);
 
                 if (flush_on_commit) {
-                        btrfs_start_delalloc_inodes(root);
-                        ret = btrfs_wait_ordered_extents(root, 0);
+                        btrfs_start_delalloc_inodes(root, 1);
+                        ret = btrfs_wait_ordered_extents(root, 0, 1);
                         BUG_ON(ret);
                 } else if (snap_pending) {
-                        ret = btrfs_wait_ordered_extents(root, 1);
+                        ret = btrfs_wait_ordered_extents(root, 0, 1);
                         BUG_ON(ret);
                 }
 
@@ -1070,6 +1120,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                 current->journal_info = NULL;
 
         kmem_cache_free(btrfs_trans_handle_cachep, trans);
+
+        if (current != root->fs_info->transaction_kthread)
+                btrfs_run_delayed_iputs(root);
+
         return ret;
 }
 
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 663c67404918..93c7ccb33118 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -79,6 +79,7 @@ static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
                                               struct inode *inode)
 {
         BTRFS_I(inode)->last_trans = trans->transaction->transid;
+        BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
 }
 
 int btrfs_end_transaction(struct btrfs_trans_handle *trans,
@@ -106,6 +107,10 @@ void btrfs_throttle(struct btrfs_root *root);
 int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
                                 struct btrfs_root *root);
 int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
-                                        struct extent_io_tree *dirty_pages);
+                                struct extent_io_tree *dirty_pages, int mark);
+int btrfs_write_marked_extents(struct btrfs_root *root,
+                                struct extent_io_tree *dirty_pages, int mark);
+int btrfs_wait_marked_extents(struct btrfs_root *root,
+                                struct extent_io_tree *dirty_pages, int mark);
 int btrfs_transaction_in_commit(struct btrfs_fs_info *info);
 #endif
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 4edfdc2acc5f..4a9434b622ec 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -542,8 +542,8 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 
         saved_nbytes = inode_get_bytes(inode);
         /* drop any overlapping extents */
-        ret = btrfs_drop_extents(trans, root, inode,
-                         start, extent_end, extent_end, start, &alloc_hint, 1);
+        ret = btrfs_drop_extents(trans, inode, start, extent_end,
+                                 &alloc_hint, 1);
         BUG_ON(ret);
 
         if (found_type == BTRFS_FILE_EXTENT_REG ||
@@ -930,6 +930,17 @@ out_nowrite:
         return 0;
 }
 
+static int insert_orphan_item(struct btrfs_trans_handle *trans,
+                              struct btrfs_root *root, u64 offset)
+{
+        int ret;
+        ret = btrfs_find_orphan_item(root, offset);
+        if (ret > 0)
+                ret = btrfs_insert_orphan_item(trans, root, offset);
+        return ret;
+}
+
+
 /*
  * There are a few corners where the link count of the file can't
  * be properly maintained during replay.  So, instead of adding
@@ -997,9 +1008,13 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
         }
         BTRFS_I(inode)->index_cnt = (u64)-1;
 
-        if (inode->i_nlink == 0 && S_ISDIR(inode->i_mode)) {
-                ret = replay_dir_deletes(trans, root, NULL, path,
-                                         inode->i_ino, 1);
+        if (inode->i_nlink == 0) {
+                if (S_ISDIR(inode->i_mode)) {
+                        ret = replay_dir_deletes(trans, root, NULL, path,
+                                                 inode->i_ino, 1);
+                        BUG_ON(ret);
+                }
+                ret = insert_orphan_item(trans, root, inode->i_ino);
                 BUG_ON(ret);
         }
         btrfs_free_path(path);
@@ -1587,7 +1602,6 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
                 /* inode keys are done during the first stage */
                 if (key.type == BTRFS_INODE_ITEM_KEY &&
                     wc->stage == LOG_WALK_REPLAY_INODES) {
-                        struct inode *inode;
                         struct btrfs_inode_item *inode_item;
                         u32 mode;
 
@@ -1603,31 +1617,16 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
                                              eb, i, &key);
                         BUG_ON(ret);
 
-                        /* for regular files, truncate away
-                         * extents past the new EOF
+                        /* for regular files, make sure corresponding
+                         * orhpan item exist. extents past the new EOF
+                         * will be truncated later by orphan cleanup.
                          */
                         if (S_ISREG(mode)) {
-                                inode = read_one_inode(root,
-                                                       key.objectid);
-                                BUG_ON(!inode);
-
-                                ret = btrfs_truncate_inode_items(wc->trans,
-                                        root, inode, inode->i_size,
-                                        BTRFS_EXTENT_DATA_KEY);
+                                ret = insert_orphan_item(wc->trans, root,
+                                                         key.objectid);
                                 BUG_ON(ret);
-
-                                /* if the nlink count is zero here, the iput
-                                 * will free the inode.  We bump it to make
-                                 * sure it doesn't get freed until the link
-                                 * count fixup is done
-                                 */
-                                if (inode->i_nlink == 0) {
-                                        btrfs_inc_nlink(inode);
-                                        btrfs_update_inode(wc->trans,
-                                                           root, inode);
-                                }
-                                iput(inode);
                         }
+
                         ret = link_to_fixup_dir(wc->trans, root,
                                                 path, key.objectid);
                         BUG_ON(ret);
@@ -1977,9 +1976,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 {
         int index1;
         int index2;
+        int mark;
         int ret;
         struct btrfs_root *log = root->log_root;
         struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
+        unsigned long log_transid = 0;
 
         mutex_lock(&root->log_mutex);
         index1 = root->log_transid % 2;
@@ -1994,12 +1995,13 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
         if (atomic_read(&root->log_commit[(index1 + 1) % 2]))
                 wait_log_commit(trans, root, root->log_transid - 1);
 
-        while (root->log_multiple_pids) {
+        while (1) {
                 unsigned long batch = root->log_batch;
-                mutex_unlock(&root->log_mutex);
-                schedule_timeout_uninterruptible(1);
-                mutex_lock(&root->log_mutex);
-
+                if (root->log_multiple_pids) {
+                        mutex_unlock(&root->log_mutex);
+                        schedule_timeout_uninterruptible(1);
+                        mutex_lock(&root->log_mutex);
+                }
                 wait_for_writer(trans, root);
                 if (batch == root->log_batch)
                         break;
@@ -2012,7 +2014,16 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
                 goto out;
         }
 
-        ret = btrfs_write_and_wait_marked_extents(log, &log->dirty_log_pages);
+        log_transid = root->log_transid;
+        if (log_transid % 2 == 0)
+                mark = EXTENT_DIRTY;
+        else
+                mark = EXTENT_NEW;
+
+        /* we start IO on  all the marked extents here, but we don't actually
+         * wait for them until later.
+         */
+        ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark);
         BUG_ON(ret);
 
         btrfs_set_root_node(&log->root_item, log->node);
@@ -2023,9 +2034,9 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
         root->log_start_pid = 0;
         smp_mb();
         /*
-         * log tree has been flushed to disk, new modifications of
-         * the log will be written to new positions. so it's safe to
-         * allow log writers to go in.
+         * IO has been started, blocks of the log tree have WRITTEN flag set
+         * in their headers. new modifications of the log will be written to
+         * new positions. so it's safe to allow log writers to go in.
          */
         mutex_unlock(&root->log_mutex);
 
@@ -2046,6 +2057,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
         index2 = log_root_tree->log_transid % 2;
         if (atomic_read(&log_root_tree->log_commit[index2])) {
+                btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
                 wait_log_commit(trans, log_root_tree,
                                 log_root_tree->log_transid);
                 mutex_unlock(&log_root_tree->log_mutex);
@@ -2065,14 +2077,17 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
          * check the full commit flag again
          */
         if (root->fs_info->last_trans_log_full_commit == trans->transid) {
+                btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
                 mutex_unlock(&log_root_tree->log_mutex);
                 ret = -EAGAIN;
                 goto out_wake_log_root;
         }
 
         ret = btrfs_write_and_wait_marked_extents(log_root_tree,
-                                &log_root_tree->dirty_log_pages);
+                                &log_root_tree->dirty_log_pages,
+                                EXTENT_DIRTY | EXTENT_NEW);
         BUG_ON(ret);
+        btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
 
         btrfs_set_super_log_root(&root->fs_info->super_for_commit,
                                 log_root_tree->node->start);
@@ -2092,9 +2107,14 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
          * the running transaction open, so a full commit can't hop
          * in and cause problems either.
          */
-        write_ctree_super(trans, root->fs_info->tree_root, 2);
+        write_ctree_super(trans, root->fs_info->tree_root, 1);
         ret = 0;
 
+        mutex_lock(&root->log_mutex);
+        if (root->last_log_commit < log_transid)
+                root->last_log_commit = log_transid;
+        mutex_unlock(&root->log_mutex);
+
 out_wake_log_root:
         atomic_set(&log_root_tree->log_commit[index2], 0);
         smp_mb();
@@ -2133,12 +2153,12 @@ int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root)
 
         while (1) {
                 ret = find_first_extent_bit(&log->dirty_log_pages,
-                                    0, &start, &end, EXTENT_DIRTY);
+                                0, &start, &end, EXTENT_DIRTY | EXTENT_NEW);
                 if (ret)
                         break;
 
-                clear_extent_dirty(&log->dirty_log_pages,
-                                   start, end, GFP_NOFS);
+                clear_extent_bits(&log->dirty_log_pages, start, end,
+                                  EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS);
         }
 
         if (log->log_transid > 0) {
@@ -2862,6 +2882,21 @@ out:
         return ret;
 }
 
+static int inode_in_log(struct btrfs_trans_handle *trans,
+                 struct inode *inode)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        int ret = 0;
+
+        mutex_lock(&root->log_mutex);
+        if (BTRFS_I(inode)->logged_trans == trans->transid &&
+            BTRFS_I(inode)->last_sub_trans <= root->last_log_commit)
+                ret = 1;
+        mutex_unlock(&root->log_mutex);
+        return ret;
+}
+
+
 /*
  * helper function around btrfs_log_inode to make sure newly created
  * parent directories also end up in the log.  A minimal inode and backref
@@ -2901,6 +2936,11 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
         if (ret)
                 goto end_no_trans;
 
+        if (inode_in_log(trans, inode)) {
+                ret = BTRFS_NO_LOG_SYNC;
+                goto end_no_trans;
+        }
+
         start_log_trans(trans, root);
 
         ret = btrfs_log_inode(trans, root, inode, inode_only);
diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h
index d09c7609e16b..0776eacb5083 100644
--- a/fs/btrfs/tree-log.h
+++ b/fs/btrfs/tree-log.h
@@ -19,6 +19,9 @@
 #ifndef __TREE_LOG_
 #define __TREE_LOG_
 
+/* return value for btrfs_log_dentry_safe that means we don't need to log it at all */
+#define BTRFS_NO_LOG_SYNC 256
+
 int btrfs_sync_log(struct btrfs_trans_handle *trans,
                    struct btrfs_root *root);
 int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root);
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 7eda483d7b5a..41ecbb2347f2 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1135,7 +1135,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
                 root->fs_info->avail_metadata_alloc_bits;
 
         if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) &&
-            root->fs_info->fs_devices->rw_devices <= 4) {
+            root->fs_info->fs_devices->num_devices <= 4) {
                 printk(KERN_ERR "btrfs: unable to go below four devices "
                        "on raid10\n");
                 ret = -EINVAL;
@@ -1143,7 +1143,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
         }
 
         if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) &&
-            root->fs_info->fs_devices->rw_devices <= 2) {
+            root->fs_info->fs_devices->num_devices <= 2) {
                 printk(KERN_ERR "btrfs: unable to go below two "
                        "devices on raid1\n");
                 ret = -EINVAL;
@@ -1434,8 +1434,8 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
                 return -EINVAL;
 
         bdev = open_bdev_exclusive(device_path, 0, root->fs_info->bdev_holder);
-        if (!bdev)
-                return -EIO;
+        if (IS_ERR(bdev))
+                return PTR_ERR(bdev);
 
         if (root->fs_info->fs_devices->seeding) {
                 seeding_dev = 1;
@@ -2209,7 +2209,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
                 max_chunk_size = 10 * calc_size;
                 min_stripe_size = 64 * 1024 * 1024;
         } else if (type & BTRFS_BLOCK_GROUP_METADATA) {
-                max_chunk_size = 4 * calc_size;
+                max_chunk_size = 256 * 1024 * 1024;
                 min_stripe_size = 32 * 1024 * 1024;
         } else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
                 calc_size = 8 * 1024 * 1024;
@@ -2538,6 +2538,11 @@ int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset)
         if (!em)
                 return 1;
 
+        if (btrfs_test_opt(root, DEGRADED)) {
+                free_extent_map(em);
+                return 0;
+        }
+
         map = (struct map_lookup *)em->bdev;
         for (i = 0; i < map->num_stripes; i++) {
                 if (!map->stripes[i].dev->writeable) {
@@ -2649,8 +2654,10 @@ again:
         em = lookup_extent_mapping(em_tree, logical, *length);
         read_unlock(&em_tree->lock);
 
-        if (!em && unplug_page)
+        if (!em && unplug_page) {
+                kfree(multi);
                 return 0;
+        }
 
         if (!em) {
                 printk(KERN_CRIT "unable to find logical %llu len %llu\n",
diff --git a/fs/btrfs/xattr.c b/fs/btrfs/xattr.c
index b0fc93f95fd0..193b58f7d3f3 100644
--- a/fs/btrfs/xattr.c
+++ b/fs/btrfs/xattr.c
@@ -85,22 +85,23 @@ out:
         return ret;
 }
 
-int __btrfs_setxattr(struct inode *inode, const char *name,
-                            const void *value, size_t size, int flags)
+static int do_setxattr(struct btrfs_trans_handle *trans,
+                       struct inode *inode, const char *name,
+                       const void *value, size_t size, int flags)
 {
         struct btrfs_dir_item *di;
         struct btrfs_root *root = BTRFS_I(inode)->root;
-        struct btrfs_trans_handle *trans;
         struct btrfs_path *path;
-        int ret = 0, mod = 0;
+        size_t name_len = strlen(name);
+        int ret = 0;
+
+        if (name_len + size > BTRFS_MAX_XATTR_SIZE(root))
+                return -ENOSPC;
 
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
 
-        trans = btrfs_join_transaction(root, 1);
-        btrfs_set_trans_block_group(trans, inode);
-
         /* first lets see if we already have this xattr */
         di = btrfs_lookup_xattr(trans, root, path, inode->i_ino, name,
                                 strlen(name), -1);
@@ -118,15 +119,12 @@ int __btrfs_setxattr(struct inode *inode, const char *name,
                 }
 
                 ret = btrfs_delete_one_dir_name(trans, root, path, di);
-                if (ret)
-                        goto out;
+                BUG_ON(ret);
                 btrfs_release_path(root, path);
 
                 /* if we don't have a value then we are removing the xattr */
-                if (!value) {
-                        mod = 1;
+                if (!value)
                         goto out;
-                }
         } else {
                 btrfs_release_path(root, path);
 
@@ -138,20 +136,45 @@ int __btrfs_setxattr(struct inode *inode, const char *name,
         }
 
         /* ok we have to create a completely new xattr */
-        ret = btrfs_insert_xattr_item(trans, root, name, strlen(name),
-                                      value, size, inode->i_ino);
+        ret = btrfs_insert_xattr_item(trans, root, path, inode->i_ino,
+                                      name, name_len, value, size);
+        BUG_ON(ret);
+out:
+        btrfs_free_path(path);
+        return ret;
+}
+
+int __btrfs_setxattr(struct btrfs_trans_handle *trans,
+                     struct inode *inode, const char *name,
+                     const void *value, size_t size, int flags)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        int ret;
+
+        if (trans)
+                return do_setxattr(trans, inode, name, value, size, flags);
+
+        ret = btrfs_reserve_metadata_space(root, 2);
         if (ret)
-                goto out;
-        mod = 1;
+                return ret;
 
-out:
-        if (mod) {
-                inode->i_ctime = CURRENT_TIME;
-                ret = btrfs_update_inode(trans, root, inode);
+        trans = btrfs_start_transaction(root, 1);
+        if (!trans) {
+                ret = -ENOMEM;
+                goto out;
         }
+        btrfs_set_trans_block_group(trans, inode);
 
-        btrfs_end_transaction(trans, root);
-        btrfs_free_path(path);
+        ret = do_setxattr(trans, inode, name, value, size, flags);
+        if (ret)
+                goto out;
+
+        inode->i_ctime = CURRENT_TIME;
+        ret = btrfs_update_inode(trans, root, inode);
+        BUG_ON(ret);
+out:
+        btrfs_end_transaction_throttle(trans, root);
+        btrfs_unreserve_metadata_space(root, 2);
         return ret;
 }
 
@@ -260,7 +283,7 @@ err:
  * attributes are handled directly.
  */
 struct xattr_handler *btrfs_xattr_handlers[] = {
-#ifdef CONFIG_BTRFS_POSIX_ACL
+#ifdef CONFIG_BTRFS_FS_POSIX_ACL
         &btrfs_xattr_acl_access_handler,
         &btrfs_xattr_acl_default_handler,
 #endif
@@ -314,7 +337,9 @@ int btrfs_setxattr(struct dentry *dentry, const char *name, const void *value,
 
         if (size == 0)
                 value = "";  /* empty EA, do not remove */
-        return __btrfs_setxattr(dentry->d_inode, name, value, size, flags);
+
+        return __btrfs_setxattr(NULL, dentry->d_inode, name, value, size,
+                                flags);
 }
 
 int btrfs_removexattr(struct dentry *dentry, const char *name)
@@ -329,10 +354,13 @@ int btrfs_removexattr(struct dentry *dentry, const char *name)
 
         if (!btrfs_is_valid_xattr(name))
                 return -EOPNOTSUPP;
-        return __btrfs_setxattr(dentry->d_inode, name, NULL, 0, XATTR_REPLACE);
+
+        return __btrfs_setxattr(NULL, dentry->d_inode, name, NULL, 0,
+                                XATTR_REPLACE);
 }
 
-int btrfs_xattr_security_init(struct inode *inode, struct inode *dir)
+int btrfs_xattr_security_init(struct btrfs_trans_handle *trans,
+                              struct inode *inode, struct inode *dir)
 {
         int err;
         size_t len;
@@ -354,7 +382,7 @@ int btrfs_xattr_security_init(struct inode *inode, struct inode *dir)
         } else {
                 strcpy(name, XATTR_SECURITY_PREFIX);
                 strcpy(name + XATTR_SECURITY_PREFIX_LEN, suffix);
-                err = __btrfs_setxattr(inode, name, value, len, 0);
+                err = __btrfs_setxattr(trans, inode, name, value, len, 0);
                 kfree(name);
         }
 
diff --git a/fs/btrfs/xattr.h b/fs/btrfs/xattr.h
index c71e9c3cf3f7..721efa0346e0 100644
--- a/fs/btrfs/xattr.h
+++ b/fs/btrfs/xattr.h
@@ -27,15 +27,16 @@ extern struct xattr_handler *btrfs_xattr_handlers[];
 
 extern ssize_t __btrfs_getxattr(struct inode *inode, const char *name,
                 void *buffer, size_t size);
-extern int __btrfs_setxattr(struct inode *inode, const char *name,
-                const void *value, size_t size, int flags);
-
+extern int __btrfs_setxattr(struct btrfs_trans_handle *trans,
+                            struct inode *inode, const char *name,
+                            const void *value, size_t size, int flags);
 extern ssize_t btrfs_getxattr(struct dentry *dentry, const char *name,
                 void *buffer, size_t size);
 extern int btrfs_setxattr(struct dentry *dentry, const char *name,
                 const void *value, size_t size, int flags);
 extern int btrfs_removexattr(struct dentry *dentry, const char *name);
 
-extern int btrfs_xattr_security_init(struct inode *inode, struct inode *dir);
+extern int btrfs_xattr_security_init(struct btrfs_trans_handle *trans,
+                                     struct inode *inode, struct inode *dir);
 
 #endif /* __XATTR__ */